Search

The aggregation of buildings is generally based on the disjoint in spatial topological relationships. However, with the diversification of data sources, the spatial topological relationships between polygons inside large buildings also become diversified and even inconsistent, making it difficult to aggregate building polygons, while the multi-scale visualization of buildings in the smart city requires rapid and accurate extraction of the contour shapes of complex buildings. For this reason, this paper proposes an aggregation method of buildings based on the shortest adjacent line between vertexes interpolated. First, recursively search for adjacent polygons based on the aggregation distance to determine the clustering regions, which is convenient for processing the map data in every region. Then, for each clustered region, encrypt vertices for each polygon using a linear interpolation method according to the order of the points on it; after that, search and construct the shortest adjacent line between each vertice and the adjacent polygon through the grid index, and delete the three types of shortest adjacent lines which are abnormal; then, merge the effective shortest adjacent lines in a certain order to form the shortest adjacent line group, and further filter out three kinds of special cases, so as to determine the aggregation area of adjacent polygons; finally, the recursive search is performed clockwise according to the minimum rotation angle, and the contour of polygons including the inner ring and the outer ring is extracted. Experiments show that this method can quickly detect the aggregation area between the adjacent polygons. Experiments in this paper show that this method can quickly detect the adjacent aggregation areas between building polygons and accurately extract the outer contour lines of complex building entities in real-time. The overall contours extracted are close to the actual situation, which provides conditions for building visualization.

In order to effectively solve the trust problem of both parties in the process of geographic data transaction, this paper proposes a geographic data transaction certificate and copyright protection model combining zero-watermark, IPFS and smart contract technology. The model uses the IPFS system to build a new mechanism for zero-watermark registration, which can eliminate the dependence on third-party IPR institutions. Through the smart contract design, the zero-watermark information is registered, and the transaction information is stored permanently in the blockchain to realize the time-stamp authentication of zero-watermark registration. The tamper-resistant of blockchain and the automatic execution mechanism of smart contract in the model enable both parties to reach a consensus and achieve mutual trust without the supervision of a third party. Based on this model, this paper uses "Ant OpenChain" as a smart contract development platform to develop a transaction and copyright protection system for geographic data. With this system, both sides of the transaction can call the smart contract to obtain the transaction certificate from the blockchain. Then, the watermark information is extracted by the zero-watermark downloaded from the IPFS network. Finally, the data copyright can be confirmed and the source can be traced effectively for protecting rights or solving disputes by combining the transaction certificate and watermark information to confirm. Experiments show that the proposed method can protect the rights and interests of both parties and protect the security of transaction data simultaneously.

The selection of road network is a fuzzy decision-making process. Aiming at problems such as noise, conflict and human decision-making in the cartographic generalization algorithm based on general case reasoning, this paper proposes a road network selection method which uses ontology to organize cases and conduct knowledge reasoning. In this paper, we extract expert cases from series scale map and describe them formally according to ternary model. To realize knowledge reasoning, these cases are used to build a domain ontology repository, reorganizing their representation format. Then, ontology rule extension helps to identify and eliminate noise and conflict cases. Additionally, with methods of semantic mapping and graph query, the semantic and geometric features of the road network to be selected are associated with the ontology repository to realize road network selection automatically.

Whether the location is accurately described or not directly affects the quality of service of LBS. Relative location description is a common way of location information exchange within the text big data, and its core element lies in the selection of reference landmarks. Accurate and effective evaluation of landmark influence is an important foundation for correct selection of a reference landmark. The models involved in the current research focus on the impact on the inherent characteristics of landmarks. However, non-isolated landmarks are always subject to the dual constraints on their internal characteristics and external landmarks at the same time. Therefore, the influence of distance between landmarks can not be ignored in actual scenes. This paper evaluates the individual influence of landmarks from the aspects of cognition, vision and scene, and calculates the relative influence of each landmark on the point to be located by using the distance between the point and the landmark. A landmark influence evaluation model considering distance constraints is proposed to solve above problems. Finally, take multiple adjacent landmarks in the surrounding area of Bayi square in Nanchang as an example, an on-site questionnaire surveys was performed to verify the effectiveness of the evaluation model. The results show that the model proposed in the paper has a good performance and most recommended landmarks could meet people's psychological expectations than the existing models.

Extracting accurate road curb is a crucial task for driverless vehicles. However, existing road curb points extraction methods are not robust for sparse 16-ray LiDAR data. This paper presents a road curb points extraction algorithm that combines multiscale spatial features and measuring distance. The points outside the road areas are firstly removed by adopting the random sample consensus (RANSAC) algorithm, then most of the road surface points are removed by judging the horizontal and vertical continuity between points in the same laser beam. According to the measurement model of the road curb points, if the measured distance of the reserved points is within a reasonable distance and the angle between the two horizontal vectors starting with the point is larger than a certain threshold, the point will be identified as the road curb point. Experiments show that the road curb points extraction method proposed in this paper performs better than the other methods under 16-ray LiDAR dataset and meets the real-time requirements for environmental perception of driverless vehicles.

Road intersections are important components of a road network, which are not only numerous and diverse in shape, but also complex in structure and different in size. It is difficult to recognize comprehensive and accurate road junctions based on single data source, as its limited describe information. To this end, this paper designs a multiple integration method to identify road intersections from vehicle trajectories and remote sensing images. Firstly, based on the unsupervised idea, a method combining morphological processing, density peak clustering and tensor voting is proposed to extract the seed intersections, which is regarded as a small sample set. Based on it two intersection classifiers based on deep convolution network and oriented to vehicle trajectories and remote sensing images are constructed by using collaborative training mechanism, and finally, the advantages of the two models are combined to form an integrated classification model of road intersections. In this paper, a semi supervised intersection extraction technology is proposed by fusing the complementary description features of vehicle trajectories and remote sensing images on multiple levels, which can effectively identify complex and diverse road intersections without manual labeling. Experiments based on Wuhan taxi trajectories and remote sensing images show that the accuracy of this method is more than 93% and the recall rate is 87% without manually labeled samples.

Visual saliency modeling of driving scenes is an important research direction in intelligent driving, especially in the areas of assisted driving and automatic driving. The existing visual saliency modeling methods for static and virtual scenes cannot adapt to the real-time, dynamic and task-driven characteristics of road scenes in real driving environments. Building a visual saliency model of dynamic road scenes in real driving environments is a challenge for current research. Starting from the characteristics of driving environment and driver's visual cognitive law, this paper extracts low-level visual features, high-level visual features and dynamic visual features of road scenes, and combines two influencing factors of speed and road curvature to build a visual saliency calculation model of driving scenes based on logistic regression model (LR). In this paper, the AUC value is used to evaluate the model, and the results showed an accuracy of 90.43%, which is significant advantage over traditional algorithms.

From the automatic driving vehicle to the intelligent vehicle, and then to intelligent connected vehicle, the intelligent driving vehicle technology and industry has developed rapidly, surveying and remote sensing technology have played an important supporting role. Firstly, this paper introduces the development of intelligent vehicles at home and abroad and their differences from conventional vehicles. Based on the perspective of three important waves of scientific and technological development, this paper comparatively analyzes and summarizes the development process and core technology driving forces of autonomous driving and surveying remote sensing. The intelligent driving vehicle's key technologies are then introduced from four aspects: top-level planning and policy environment, environment perception and computing decision-making vehicle key technology, high definition map and navigation positioning basic support key technology, and vehicle road collaborative information interaction key technology. Several advanced navigation and positioning technology achievements, such as satellite-based navigation enhancement, high precision position and attitude measurement, multi-source fusion perception, are taken as examples to demonstrate that surveying and mapping can enable intelligent driving of a single car. By introducing mobile measurement and crowdsourcing high precision map-making technology, the key supporting role of surveying and remote sensing for intelligent driving is expounded. This paper takes the construction of space-based information real-time service system as an example to describe the development trend of how surveying, mapping, and remote sensing will serve the "person vehicle road cloud" of intelligent connected vehicles in the future.Finally, the paper summarizes and analyzes the problems that need to be solved in developing intelligent (connected) vehicles and the challenges that need to be addressed in the near future.

Accurate monitoring of surface water is an important basic application of remote sensing. The principle of optical remote sensing water extraction is based on different ground features having different spectral reflection characteristics. Some ground features (ice, snow, shadows, clouds) have similar reflection characteristics to water bodies, which leads to the failure of extraction and classification. Aiming at the problem of misclassification and omission of traditional water body index in water body extraction, this paper proposes a normalized difference multi-band water index model. This paper uses two experiments to test the stability of the new index. The area of experiment 1 is the Linzhi area of Tibet. The data source is Landsat 8 and Sentinel 2 satellite images in the same time phase. The experimental results verify the ability of the new index to suppress snow and ice. The Kappa coefficient of the new index is 0.86, the overall accuracy is 0.93, and the misclassification error is 0.03, the omission error is 0.12, the drawing accuracy is 0.97, and the producer accuracy is 0.88, which are better than the existing index. The data source of experiment 2 was GF-1, and Hong Kong Disneyland was used as the experimental area. Experiment 3 extracted water bodies in multiple regions and verified the stability of the water body index in this paper. Water extraction was performed in an environment with a small amount of clouds, which proved that the new index can suppress clouds and their shadows. This paper uses multi-source optical remote sensing image to verify the feasibility of the new index. Without additional auxiliary data, the influence of snow, cloud and shadow can be eliminated, and the water can be more effectively and automatically extracted, which can be extended to coastal resource research, glacier change, inland lake change and other fields.

Aerosol optical depth (AOD) is a basic parameter of total aerosol content, which plays an important role in the study of atmospheric air quality. In order to explore the impact of different types of AOD on air quality, this paper proposes two adaptive AOD prediction methods based on GNSS PWV. The proposed method considers the temporal autocorrelation of AOD between adjacent epochs, and the model coefficients can be updated adaptively. One method is to model the 550 nm AOD directly based on GNSS PWV, which is called TAF (total AOD forecast) model. Another AOD modeling method takes into account the sensitivity of five different types of AOD to PWV, referred to it simply as FTAF(five type based AOD forecast) model. The model first establishes the functional relationship between PWV and five types of AOD. Secondly, according to the relationship between 550 nm AOD and five types of AOD, the weight of different types of AOD in 550 nm AOD was determined. Finally, PWV is used to predict five types of AOD, and the final 550 nm AOD is obtained by weighted average. 16 GNSS stations in Beijing, Tianjin, Hebei region are selected to verify the accuracy of the proposed model. The results show that the two 550 nm AOD prediction models have high accuracy, and the FTAF model is better than the TAF model. The AOD prediction model proposed in this paper can effectively apply the tropospheric parameters retrieved from GNSS to remote sensing monitoring of atmospheric environment, which provides a new idea for the study of atmospheric environmental quality.

With the development of Global Navigation Satellite Systems (GNSS), the multipath effect that was once regarded as one error source has been proven can be used to monitor reflecting surface parameters such as water level, snow depth, vegetation water content, and soil moisture using a technology named GNSS-interferometry reflectomety (GNSS-IR). In order to expand monitoring objects and applications of GNSS-IR, this paper proposed the theory and method to retrieve wind speed using ground-based GNSS-IR. The theory includes the signal-to-noise ratio (SNR) multipath oscillation theory, the scattering model theory, and wave spectrum theory. This paper simulated the mathematical model between the SNR cutoff elevation angle and the wind speed, and used the wavelet analysis to extract SNR cutoff elevation angle and a unified method to unify reference of cutoff elevation angle. Data of HKQT station, located in Hong Kong, during the period of "Mangkhut" and "Hato" were used. The results showed that the variations of GPS L5 SNR cutoff elevation angles had a good corresponding relationship with the measured wind speeds, and the correlation coefficient reached 70%~85%. Also, the variations of cutoff elevation angles can describe the change of wind speed from low wind speed to high wind speed and there was a specific mathematical relationship between them. Finally, this paper discussed two difficulties to overcome in the GNSS-IR wind speed retrieval technology.

Urban traffic efficiency is one of the key factors affecting urban productivity and is also a crucial topic in the process of smart city construction. With the development of computer technology, artificial intelligence, especially reinforcement learning, plays an increasingly important role in traffic signal control. Currently, traffic signal control based on reinforcement learning is mainly used for the optimization for simple scenarios, such as single road intersection or urban arterial road, not yet for regional coordinated control on an urban geographic road network. This paper is motivated to fill this gap by proposing a two-layered agent cooperative control approach based on reinforcement learning. The first layer implements a coarse-tuning training at a single intersection, where the agents make the single intersection non-blocking by observing the queue length for each lane; In the second layer, the coarse-tuning-trained agent models are put into the geographic network to execute the cooperative fine-tuning training at multiple intersections. This paper conducts the optimization-orientated traffic coordination through a case study of a middle school area in Ningbo. The results show that our control approach is superior to the traditional fixed timing scheme in terms of the passage efficiency.

The ranging accuracy of spaceborne laser altimeter is one of the main sources that affect the geometric calibration and processing accuracy of the laser.Aiming at the issues of laser ranging extraction lower accuracy and the stability, caused by the quantization error of the full-waveform spaceborne laser analog signal after digital processing, so a sliding window Gaussian fitting algorithm for ranging error suppression of full-waveform spaceborne laser is proposed.This method uses a sliding window to eliminate noise points near the peak, and optimizes the peak value of the waveform to extract a more accurate ranging value.Finally, the GF-7 laser altimeter was used as the test object, the relative and absolute accuracy of laser elevation were compared and verified by using the ice surface, inland lake and the flat ground.The results indicate that, the laser ranging accuracy of the paper algorithm is improved by 7.5 cm compared with the general peak method. Based on the ranging value of the paper method, the relative accuracy of laser elevation is improved by 4.3 cm. The absolute accuracy of laser elevation is increased by 4.5 cm verified by airborne LiDAR point cloud data. Which fully shows that the method in this paper can be used as a method to reduce effectively the random error of laser ranging, and provides an indispensable foundation for the sub-meter elevation measurement accuracy of the GF-7 satellite.

The dynamic precision engineering survey is not only used in the construction but also in monitoring the dynamic change during usage. With the rapid popularization of various sensors and automated mobile platform, such as intelligent vehicles, unmanned aerial vehicle (UAV), unmanned ship, robots etc., the engineering survey is upgraded in an automated, dynamic and intelligent way. It is now armed with the ability to survey in movement. Surveying devices, including mature surveying robots, mobile surveying vehicle, etc., have been proposed. Another way is to assemble devices delicately for specific applications, which can achieve dynamic and precise measurements of position, posture, internal deform and external shape. Focusing on the intelligent device with integrated sensors and specific applications, this paper puts forward the integration principle of multiple sensors and the advanced data processing methods for intelligent engineering surveying. This paper also demonstrates several applications of these devices, such as internal deformation monitoring of the rockfill dam, damage monitoring of the urban underground water pipelines, surveying of the road damages and the damage detection on the south-to-north water diversion project (SNWDP), to show different requirements and solutions in different applications.

This paper proposes the concept of surveying robot, focuses on analyzing the framework and key technologies of the surveying robot, covers the cutting-edge research fields of surveying robots such as positioning and mapping, scene perception, path planning, and multi-machine collaboration. This paper also summarizes the latest research progress, analyzes the current technical problems and describes the development opportunities of surveying robot.

With the rapid development of geospatial science, artificial intelligence, and high-performance computing, geospatial artificial intelligence has become the major technique for processing and analyzing geospatial big data, and will be widely used in the scientific research and engineering application of earth science, spatial cognition, and smart city. Geospatial intelligence, as an inter-discipline between geospatial science and artificial intelligence, is driven by dual disciplines. At present, it has made important progress in hardware research, system development, data and model sharing, services and applications, and is also facing new challenges. This paper first describes the conceptual evolution of geospatial artificial intelligence and lists the frameworks of some technical systems. The paper then reviews the state-of-the-art in the scientific research and typical application domains. The problems and challenges facing geospatial intelligence are analyzed. Finally, some future prospects and trends are presented.

Geographic knowledge service (GKS) is widely believed to be the intelligent successor to the current geographic information service (GIS). The need for GKS is becoming ever pressing due to the increasing severity of the information explosion problem. This paper clarifies the confusions in the connotation of GKS, its relationship with GIS, and its requirements for formal representation and intelligent processing of geographic knowledge. After an in-depth analysis of the latest breakthroughs made in artificial intelligence (AI), it is argued that the state-of-the-art AI provides a promising basis of cognitive intelligence for advancing GKS development. Emphasizing the spatio-temporal characteristic of geographic knowledge, this paper identifies three main categories of research issues in GKS development, i.e. those of the knowledge engineering approach to geographic modeling, the cognitively intelligent approach to geographic analysis and the context-aware computing approach to service provision. A graded strategy for advancing GKS is then suggested and a roadmap of GKS development is envisioned.

Carrying out an integrated survey, monitoring and evaluation of natural resources, accurately understanding the status and changes of various natural resources in China, is the scientific basis for territorial and spatial plans, and gradually realizing the overall protection, restoration, and comprehensive management of landscapes (including mountains, forests, fields, lakes and grasses), ensuring national ecological security. At present, the extraction of natural resource features based on remote sensing images mainly relies on visual interpretation via man-machine interaction and field spot verification. It needs high labor intensity, and production efficiency is low. The results are also highly affected by man-induced factor, which can no longer adapt to the requirements for integrated investigation and monitoring of all features of natural resources. This paper conforms to the emerging direction of the research development with artificial intelligence collaboration. Firstly, this paper reviews the main research progress of deep learning technology and its application in the field of remote sensing image intelligent extraction systematically, and analyzes its limitations, then it reviews the main research status of man-machine collaboration. Afterward, Starting from the characteristics of natural resource features, presents a technical framework of "intelligent background computing+intelligent engine+man-machine interface" for man-machine collaborative intelligent extraction. The key technologies that need to be overcome are described. At last, the idea of creating cloud platform for feature extraction are discussed. This paper aims to provide a new AI method for intelligent extraction and improve the automation and intelligence level of natural resource feature extraction.

In this paper, we reviewed the operational topographic surveying and mapping capability, i. e., plane and height accuracy, that synthetic aperture radar (SAR) satellites achieve. The plane accuracy of TerraSAR-X, COSMO-SkyMed and Tianhui-2 meets the need of 1:50 000 digital orthophoto map standard. The state-of-the-art plane accuracy is 2 m which is achieved by TerraSAR-X. Meanwhile, the accuracy of operational ground control point production provided by TerraSAR-X is 3 cm. We reviewed the error sources and the calibration methods for centimetric SAR image location. The height accuracy of TanDEM-X and Tianhui-2 meets the needs of 1:50 000 digital elevation model standard. The state-of-the-art height accuracy is 0.88 m which is achieved by TanDEM-X if the dense vegetation areas and the snow-and-ice covered areas are exclusively included in the statistical results. We reviewed the error sources and the calibration methods for height measurement without ground control information. The paper is expected to provide references for the geolocation and height measurement of the domestic SAR satellites.

GNSS three-dimensional water vapor tomography technology is used to obtain all-weather atmospheric water vapor distribution information with high spatial and temporal resolution. At present, the observation data of the tomographic model considering only the top signal have low utilization rate and large grid space rate, and the vertical constraint equation adopted is less consistent with the actual water vapor distribution. This paper based on the function fitted with sounding information to establish a vertical constraints, designs and implements a tomographic method considering boundary signals, uses the observation data of Hong Kong CORS network and radio sounding products. The improvement of the results of water vapor tomography by introducing boundary signals and for vertical constraint equation are analyzed in detail. The results show that the addition of boundary signals increases the observation equation by 51.9%, decreases the grid space rate significantly, and reduces the root-mean-square error of the solution results by 12.1%. Compared with the traditional constraint equation based on the exponential decline characteristic in the vertical direction, the root-mean-square error of the vertical constraint solution adopted in this paper is reduced by 5.7%.

The Bootstrap resampling method is introduced to the nonlinear theory for solving the precision estimation in this paper. By resampling the original sample observation data or the residuals of the dependent variable to obtain Bootstrap samples instead of the complex derivative calculations, the complete algorithms of Bootstrap method for solving the problem of nonlinear accuracy evaluation are given. Aiming at the equal probability resampling of model stochastic variable, this paper obtains the empirical distribution function of the stochastic variable in the sampling process, proposes the weighted sampling strategy, and gives the detailed calculation steps of the improved method for the accuracy evaluation. The results of experiments show that the Bootstrap method based on the resampling observations and the Bootstrap method based on the resampling residuals have stronger applicability, and can obtain more reasonable parameter standard deviations than the approximate function method and Jackknife method. Furthermore, the weighted resampling Bootstrap method based on the resampling observations and the weighted resampling Bootstrap method based on the resampling residuals can obtain more accurate precision information with extensive advantages. Those which verified the feasibility and effectiveness of using Bootstrap method and the improved algorithms proposed in this paper for precision estimation of nonlinear adjustment.

There are significant nonlinear intensity differences between multi-modal images. Moreover, the noise in these images will cause image degradation. Therefore, the automatic registration of multi-modal images is a challenging task. To address the two problems, this paper proposes a multi-modal image automatic registration method, which is divided into two stages: pre-registration and fine registration. In the pre-registration stage, an improved SIFT algorithm is used to roughly align multi-modal images. In the fine registration stage, the block Harris detector is first used to extract evenly distributed feature points on the pre-registered reference image. Then, the structure information in the multi-modal images is captured by the anisotropic structure tensor to construct a feature descriptor, which is robust to noise. Furthermore, a similarity criterion named TOMI (tensor orientation and mutual information) is proposed combining the tensor orientation parallelism and gradient mutual information. Finally, Multi-modal images (including Optical, LiDAR, SAR, and Map data) are used to evaluate the proposed algorithm. The experimental results show that the method proposed in this paper is robust to nonlinear intensity differences and noise, and the matching effect is superior.

Provenance is an important research issue of “Neogeography”, and it plays a significant role in judging whether geospatial data is reliable or not. In a distributed collaborative environment, geospatial data provenance also faces reliability issues. Blockchain technology has gained rapid development in recent years due to its characteristics of credibility, transparency, and non-tampering. It provides a new solution to the reliability management of provenance records. However, how to use the blockchain to organize and store a large number of geospatial data provenance with different granularity levels remains a challenge. Therefore, taking vector data as the research object, this paper discusses the requirements in designing the structure of provenance blockchain, and proposes a method of organizing on-chain provenance information at different levels based on Merkle Patricia tree (MPT). Further, an algorithm that is suitable for the verification of vector provenance information is provided.This paper develops a vector data provenance blockchain prototype system, and conducts verification experiments using the vector provenance information. The experiments demonstrate that MPT can achieve higher performance than binary Meckel tree in vector data provenance verification.

Spatial cognition affects human spatial behaviors. The study of cognitive process is of great significance in map and navigation design, geography teaching and so on. Traditional cognitive experiments mostly use maps (including paper maps and electronic maps, street view maps, etc.) and real scenes as experimental environments to study the psychological process of subjects’ reading maps and the guidance effect of maps in real geographic environment. This paper puts forward the concept of earth view, that is, taking the third person reading scene of Google Earth as the experimental environment, studying how the candidates orientate and locate the places of countries or regions and search for the target on the Earth without fixed axis through certain operations on Google Earth, and analyzing the cognitive process of the whole earth sphere. By visualizing the process of reading the Earth with the gaze point data recorded by eye tracker and analyzing the process of thinking, it is found that on the earth, the target searching process, which combines relative position judgment and orientation positioning, is the comprehensive presentation of construction process about the candidates’ Earth view.

At present, eye tracking technology has been widely used in human-computer interaction, user behavior recognition and prediction, but how to automatically identify user’s eye movement behavior in map reading is still a challenge. This paper proposed a method based on the naive Bayesian classification model to identify the users’ behavior when performing linear feature recognition. We first conducted an eye tracking experiment to acquire users’ eye movement dataset of map reading. Then we extracted and discretized 250 eye movement features involved in the algorithm, and used minimum redundancy maximum relevance algorithm to further select the features. The results show that when the attribute selection method is m=5 using mutual information quotient, the classification accuracy is 78.27%. And when using mutual information difference and m=4, the classification accuracy is 77.01%.We suggested that the proposed method can effectively identify the first elements in the map using eye movement data. This study explores the interaction technology by combining the eye tracking, laying the foundation for the future of designing gaze-controlled interactive map. The proposed method based on naive Bayesian model in this paper is comparable to the existing methods. In addition, the execution efficiency of the model is greatly improved due to the reduction in the number of features. The eye-track recognition algorithm of map reading behavior proposed in this study lays a foundation for future gaze-controlled interactive map research.

Focused on the geographic spatial cognition, this paper introduces the origination, the current situations, the four classical geographic spatial cognition theories and the difference between two factions. In the following, for the important role of map in spatial cognition, the function and research methods are concluded. Especially, the eye-tracking technology are paid more attention to, for its advantages in evaluation and interaction. Eventually, three developing directions and the characters of geographic spatial cognition in new AI era are discussed. With a focus on theoretical issues of geospatial cognition, this paper addresses at first their origin in psychology and the state of the art. The classical theories of geospatial cognition and their impacts are then introduced. Meanwhile, the fundamental difference between the geospatial cognition and geospatial sensing is pointed out. Although both are supported by the Artificial Intelligence (AI), the former is targeted to the understanding of how human brain works for spatial tasks, whereas the latter aims to maintain a dynamic digital twin of the geospace based on sensory technology and computer vision. The role and research methods of maps as a two-way tool for spatial cognition are discussed, with emphasis on eye-tracking experiments as well as their advantage in interactions. Three development trends of geospatial cognition in the era of AI—computational semantic cognition, cognitive mechanism of the mind and embodied cognition are summarized. The parallel development and characteristics of map-based spatial cognition in this same period are outlined.

Aiming at the ill-posed problem of multiplicative error model, this paper introduces the Tikhonov regularization method to derive the weighted least squares regularization solution. Considering the complex nonlinear relationship between parameter estimations and the observations when using weighted least squares regularization method to solve the ill-posed multiplicative error model, the scaled unscented transformation (SUT) method is used to calculate the mean value and mean square error matrix of the nonlinear function by weighted without derivation for precision estimation of ill-posed multiplicative error model. The simulated and actual examples results show that the weighted least squares regularization iterative solution proposed in this paper can effectively weaken the ill-posed model, and the precision estimation method based on SUT method can obtain more reasonable precision information than the existing methods, and has strong applicability.

Multi-temporal interferometry synthetic aperture radar (MT-InSAR) has the capability to monitor deformation with millimeter precision over wide areas. It has been widely applied for the monitoring of land subsidence. In recent years, much attention was paid on monitoring of the subsidence along large-scale man-made linear features represented by high speed railway(HSR) with MT-InSAR. This paper conducts a tentative test for applying C-band SAR data to the deformation monitoring of HSR subgrade, phase stability analysis and improved StaMPS technology were jointly used to increase the point density and stability of deformation parameter solution. In this paper, a total of 47 Sentinel-1A images spanning 21 months were processed for MT-InSAR analysis along Lian-Yan HSR, and the accuracy of linear deformation velocity and time-series deformation were evaluated respectively by the data of continuous BDS (BeiDou navigation satellite system) monitoring station. The results demonstrate that C-band Sentinel-1A data is capable of achieving millimeter accuracy in linear deformation velocity and time-series deformation, the mean root mean square error (RMSE) value of time-series displacement differences between the InSAR and the BDS is 3.8 mm, agreed well with BDS monitoring data, and the overall performance of Lian-Yan HSR is stable.

In order to solve the problem that the traditional algorithm for multibeam sounding coordinates ignores the significant impact of the difference in the position of the receiving and transmitting transducer on the deep-water sounding, and the new algorithm takes into account the difference in receiving and transmitting but low efficiency, this paper proposes a high-precision and quick algorithm for multibeam sounding coordinates that considers the propagation surface. First, based on the principle of multibeam measurement and sound ray propagation, a beam propagation surface model is proposed. Second, the beams in the sector are divided into interpolation nodes and points to be interpolated. For the former, a method based on iterative search and interpolation is proposed to avoid the establishment of a complete propagation surface. For the latter, using the interpolated node parameters, a method based on polynomial interpolation is proposed to obtain the single-way time, depression angle and azimuth angle of sound ray. Finally, all beam footprints are calculated. Experiments in shallow and medium-deep water areas show that the accuracy of the cross-line inspection method using this paper method is similar to that of Caris, and is better than the traditional algorithm. Compared with the results of Caris, the depth deviations of the same beam footprint in this paper are 0.35‰ depth and 0.11‰ depth, respectively, and the efficiency in this paper has increased by 8% and 35%, respectively.

Aiming at problems that in direction relation matrix model, recognition ability for distinguishing direction changes in the same cardinal direction is insufficient, the direction distance references for different cardinal directions are defined incompletely, and the distance calculation between arbitrarily direction relation matrices is not accurate enough, this paper proposes a direction relation two-tuple model, which combines grid-based direction relation matrix and centroid-based direction relation matrix to concern both distribution ratio variations and centroid position variations for objects, thus distinguishing the direction difference in the same cardinal direction. Meanwhile, the traditional neighborhood graph is optimized based on human spatial cognition, and a centroid direction distance reference suitable for arbitrarily direction relationships is established. Finally the Earth mover's distance (EMD) is utilized to further improve the accuracy of distance calculation between direction relation two-tuples. Experiments indicate the method is simple and feasible, the measurement results are more consistent with human cognition, and can be better applied to tasks like cartographic generalization results evaluation.

Shape similarity between buildings is useful for building matching, classification, retrieval, etc. As buildings tend to be designed with square angles, we proposed a method to measure shape similarity for buildings with multiple orthogonal features based on Needleman-Wunsch (NW) and Smith-Waterman (SW) algorithms. While encoding building shape into a sequence, adjacent two angles are taken as codes. Type, angle and length characteristics of the codes are described, and their similarity are then defined. NW and SW algorithms are applied to compute similarities between two encoding sequences of buildings. Shape-based query experiments show that proposed method in this paper can effectively measure the shape similarity between buildings. And the experimental results are also in line with human visual perception.

In urban environments, the GNSS signals could be easily blocked and reflected by buildings, which leads to low positioning accuracy and discontinuity. This paper employs k-means++, Gaussian mixed clustering (GMM) and fuzzy c-means (FCM) clustering methods to separate the line-of-sight signals (LOS), multi-path and non-line-of-sight signals (NLOS). The pseudorange residuals, signal-to-noise ratio, elevation angle, as well as pseudorange rate consistency are considered in the three methods. The performance comparisons of the three methods are carried out by static testing of GPS/BDS integrated system in known position point. The results show that FCM has the greatest silhouette coefficient and the best clustering performance. The positioning performance is therefore been greatly improve by eliminating the NLOS. The root mean squares error (RMSE) of point positioning indicates that positioning results can be improved by 50% after NLOS been eliminated by k-means++ and FCM. Compared with the supervised clustering methods, the unsupervised clustering is easy to implement without prior information, has lower computation burden and has certain advantages in improving the positioning accuracy.

Taxi trajectories not only reflect the static geometrical and topological structures of urban road network, but also carry dynamic navigation information such as turning relationship. Considering that the location, topological connection and turning rules of road intersection are the basis and key to the construction of navigable road network, this paper made full use of the spatial distribution and dynamic connection information of taxi trajectories and proposed a navigable road network generation method guided by intersection information, which adopts the research idea of "intersection location—road segment geometry—navigation attribute". Firstly, an integrated intersection recognition technology considering multi-mode features was designed, and then the random forest algorithm was introduced to explore a zero-label supervised anti-counterfeiting strategy for eliminating false intersections; on this basis, a multi-stage road segment generation method based on Delaunay triangulation network was proposed; finally, the turning relationship and one/two-way information of road segment, which are closely related to intersections, were further captured so as to derive a navigable road network. Compared with other methods, the proposed method achieves higher completeness and accuracy for the extraction of navigable road network information such as road intersection, road segment and turning relationship.

Environment perception and navigation are the core technologies for automated driving of deep space exploration rovers. Due to the special circumstances of the deep space environment, the rover's environment perception and navigation are particularly challenging comparing to the corresponding technologies in autonomous car driving on Earth. The paper presents a review of the progresses of environment perception, rover localization and path planning, from engineering application and scientific research perspectives. Moreover, the future prospects of intelligent environment perception and long-range rover navigation are discussed.

Vehicle tracking aims at estimating the target vehicle state from continuous temporal measurement. It is the core for intelligent vehicle to understand the environment and predict the targets' behaviors. LiDAR-based perception system of intelligent vehicle provides precise vehicle detection results, which are the basis of vehicle tracking. However, the tracking process suffers from issues of orientation mis-estimation and low stability of tracking trace, especially when the target vehicles are far away from LiDAR. The sparseness of point cloud at long distance is the key problem. Therefore, this paper proposes an enhanced vehicle tracking method based on the lane orientation priori from a digital map. It utilizes the OpenStreetMap digital map to fuse with the local lane markers detection results. The road model is built to obtain the constraint of lane orientation. Based on the vehicle tracking method built on extended Kalman filter, this lane orientation constraint is utilized to improve the vehicle orientation estimation. Consequently, the vehicle tracking accuracy and stability can then be reached. The tracking result indicates the multiple objects tracking accuracy can be increased by 0.33% while the average translation error can be reduced by at least 0.014 m. Moreover, the target vehicle at 60 m away from the host vehicle can be improved to reduce the error of 0.08 m.

Accurately identifying roadside objects like trees, cars, and traffic poles from mobile LiDAR point clouds is of great significance for some applications such as intelligent traffic systems, navigation and location services, autonomous driving, and high precision map. In the paper, we proposed a point-group-view network (PGVNet) to classify the roadside objects into trees, cars, traffic poles, and others, which utilize and fuse the advanced global features of multi-view images and the spatial geometry information of point cloud. To reduce redundant information between similar views and highlight salient view features, the PGVNet model employs a hierarchical view-group-shape architecture to split all views into different groups according to their discriminative level, which uses the pre-trained VGG network as the bone network. In view-group-shape architecture, global-level significant features are further generated from group descriptors with their weights. Moreover, an attention-guided fusion network is used to fuse the global features from multi-view images and local geometric features from point clouds. In particular, the global advanced features from multi-view images are quantified and leveraged as the attention mask to further refine the intrinsic correlation and discriminability of the local geometric features from point clouds, which contributions to recognize the roadside objects. We have evaluated the proposed method on five different mobile LiDAR point cloud data. Five test datasets of different urban scenes by different mobile laser scanning systems are used to evaluate the validities of the proposed method. Four accuracy evaluation metrics precision, recall, quality and Fscore of trees, cars and traffic poles on the selected testing datasets achieve (99.19%, 94.27%, 93.58%, 96.63%), (94.20%, 97.56%, 92.02%, 95.68%), (91.48%, 98.61%, 90.39%, 94.87%), respectively. Experimental results and comparisons with state-of-the-art methods demonstrate that the PGVNet model is available to effectively identify roadside objects from the mobile LiDAR point cloud, which can provide data support for elements construction and vectorization in high precision map applications.

Simultaneous localization and mapping (SLAM) has broad application prospects in many fields due to its high energy efficiency and low power consumption. However, there are still some problems in the traditional SLAM system: the key frame in the traditional visual odometer does not contain semantic information, the image information obtained by the mobile robot is relatively single, and the key frame in the actual scene always contains a large number of mismatched points and dynamic points. In response to the above problems, this paper proposes a new idea of semantic SLAM mapping technology. First, in order to find the correct and corresponding feature points, abandon the interference of dynamic points and mismatched points simultaneously, a method for judging the feature state of adjacent frames based on the Lucas-Kanade optical flow method is proposed, and this function is regarded as a new feature. The thread is added to the visual odometry part of ORB-SLAM3 to complete the optimization and improvement of part of the traditional SLAM framework. Secondly, in view of the problem that the image frame obtained by the front-end visual odometer of the traditional SLAM system does not contain any semantic information, the target detection algorithm based on YOLOV4 and the Mask R-CNN semantic segmentation algorithm fused with fully connected conditional random field CRF are used to compare ORB-SLAM3. The key frame image processing of the robot effectively improves the perception of the indoor environment of smart devices such as robots.

The floating-car GNSS trajectory data have been widely used to obtain and predict the urban traffic status in real time, with wide coverage and low deployment cost, and the result has an important supporting role for route decision-making of automatic driving and traffic management. However, the traffic information predicted by the floating-car GNSS data only contains the traffic information on each road segment, ignoring the difference of the traffic flow in different driving directions at the intersection; besides, the accuracy of the traffic information is limited by the GNSS sampling frequency. This paper proposes a turning-level traffic prediction method based on graph convolutional network and low-frequency GNSS trajectory data: first, a queuing-starting-point estimation model is proposed considering vehicle movement pattern; second, a Graph-structure of the turning connection relationship is constructed based on the dual graph theory; finally, to consider spatio-temporal pattern, a traffic prediction model is constructed based on the graph convolutional network. The experimental results show that our method can accurately obtain and predict the traffic speed and queue length at turning-level, and effectively improve the accuracy of traffic prediction by learning the spatio-temporal pattern within the Graph.

HD (high definition) maps have gradually become an indispensable part of autonomous driving, but there is no unified data model and standardize way of description, especially in the map production and data exchange stages. In response to this problem, the paper first analyzes the advantages and disadvantages of the current HD map data models such as NDS, OpenDRIVE and LaneLet, and then proposes a generalized data model for HD maps called Whu map model. In terms of the lane model, lane groups are used as data management units, which composed of one or several lanes on the same road section. And the lane is composed of the left lane divider, the right lane divider, the lane centerline, and lane attributes. Through the lane topology construction on some scenes where lane number changes, the validity and robustness of the lane model are verified. In terms of the data model of traffic facilities, the road marking model and traffic sign model are defined. The geometry, type, semantic information, and association with roads and lanes are the main contents described. Finally, through large-scale data production, compilation to NDS and OpenDRIVE format, and electronic horizon application experiments based on the ADASIS V3 protocol, the usefulness and effectiveness of the data model are verified. Whu map model can be used as a general exchange format, and can also be applied to all stages of HD map production. At the same time, it is backward compatible and easy to be extended. All of these help to realize the standardization of the HD map data model, thereby promoting the large-scale production and application of HD maps.

To improve the positioning accuracy of mobile robot, a visual SLAM algorithm based on binocular vision and inertial navigation is presented. In the front part of visual SLAM, a semi-direct binocular visual odometer combining direct method with characteristic method is presented to maintain the fast calculation speed and high accuracy of direct method. In the back-end optimization stage, the visual data and IMU data are fused together, and error functions are constructed in a sliding window in a non-linear optimization way to optimize the accuracy of pose calculation. The algorithm proposed in this paper is validated in EuRoc dataset. The results show that the positioning accuracy of the SLAM system OKVIS, ROVIO and VINS-Mono is significantly improved in both Machine Hall and Vicon Room scenarios, while maintaining high operational efficiency, compared with the open source visual inertial navigation fusion SLAM system OKVIS, ROVIO and VINS-Mono.

Given the steepness of road slope in surface mines, one unmanned truck may run at risk in unknown environment if it can't plan a proper speed in advance. Therefore, it is crucial for an autonomous vehicle to perceive an accurate value of road slope of it in real time. However, the accuracy is hard to achieve in existing methods, including global navigation satellite system (GNSS), inertial navigation system (INS) and simultaneous localization and mapping (SLAM). For GNSS or INS, they can measure a truck's angle of pitch, but this angle cannot be equal to road slope due to its large pitching and bouncing movements caused by steep and uneven roads. For the same reason, SLAM doesn't work well either. Also, it will lose efficacy as geometric features are not obvious in open-pit mine. To deal with these challenges, this paper proposes a grid Kalman road slope real-time detection (GKSRD) method. The method's input is 3D point cloud of lidar and the pitch of INS. And the method uses a 2D grid map, an iterative optimization algorithm in rectangular region of interest and the Kalman filter. Compared with methods based on GNSS or INS, this method minimizes the error of slope detection. Different from methods based on SLAM, this method does not rely on geometric features of the environment. Verified by experiments, the average error of the road slope detected by GKSRD method is less than 0.01 degree, and the maximum error is less than 0.5 degree. Hence, compared with methods based on INS or GNSS and methods based on SLAM, GKSRD method is more accurate, stable and adaptive.

Cameras can help to make low-cost positioning and environment perceiving in autopilot but dynamic objects give negative effects to visual odometry. This paper gives a model to optimize it based on semantic elements association. It uses such techniques as objects detection and semantic segmentation to identify semantic objects and distinguish dynamic semantic elements(DSE) from static ones. Then it filters bad keypoints by a dynamic feature mask in visual positioning. In practice, this proposed method detects DSE even when there are objects with duality of moving and static. In an experiment on campus roads, negative influences were found especially when a robot turning around or a moving object crossing its camera view. It showed that average accuracy in detecting DSE was 87% and the largest difference between trajectories before and after semantic association optimizing in sub-sequences was 2.463 m. Compared to ground truth, the RMSE of proposed method dropped by 38%.

To solve the problem of matching difficulty caused by the significant nonlinear grayscale differences between SAR and optical images, this paper proposes a fast matching algorithm based on image structural properties named SOFM(SAR-to-optical fast matching algorithm).The traditional methods based on image grayscale are generally difficult to resist the nonlinear grayscale differences between SAR and optical images, but the geometric constructs and shape features can exist stably among different types of images, so in our the proposed method both the magnitude and orientation information of image gradient are used to build a geometric structural feature descriptor named AWOG(angular weighted orientated gradients), then based on the template matching strategy, the sum of squared difference of the descriptors is used to define the similarity metric for matching and then the image matching function expressed in the frequency domain is given. A complete set of image matching process is established based on SOFM, and has been validated using multiple pairs of SAR and optical images, the results show that the proposed method can effectively resist the nonlinear grayscale differences between SAR and optical images, and outperforms the traditional classical image grayscale-based methods and existing image structural-based methods in matching performance and precision.

In order to solve the problem that the parameters of pulse-coupled neural network can't be adjusted adaptively in pan-sharpening image fusion, a non-subsampled shearlet transform remote sensing image fusion method based on the combination of parametric-adaptive pulse coupled neural network model and energy-attributing fusion strategy is proposed. First, the high and low frequency coefficients are obtained by extracting the Y luminance component of the multispectral image YUV color space transform and transforming it with the panchromatic image. Then, aiming at the low-frequency sub-band coefficients are fused by the EA method, the high-frequency sub-band coefficients are obtained by the PA-PCNN model to determine the optimal PCNN model, and then the high-frequency sub-band coefficients are fused; finally, the fusion image is obtained by inverse transformation of NSST and YUV. In this paper, six objective quality indexes, such as spatial frequency, relative dimensionless global error, ERGAS, correlation coefficient, visual information fidelity for fusion, gradient-based fusion performance and structural similarity index, are selected to evaluate the spectral and spatial detail information of the fused images, compared with SE, DGIF, COF and PA-PCNN fusion methods, the proposed method is validated by using multiple sets of high-and low-resolution panchromatic and multispectral remote sensing images, the results show that this method is generally superior to the traditional fusion method of panchromatic and multispectral remote sensing images in objective evaluation and visual perception.

GF-7 satellite is equipped with the first full waveform laser altimeter of China for earth observation, which is mainly used to obtain sparse elevation control points on the ground, and ultimately improve the stereo mapping accuracy of stereo images without ground control points. The laser altimetry standardization processing of GF-7 satellite is the key step of surveying and mapping application, and the generated laser altimetry standard products are the important premise of subsequent distribution and application. Based on the GF-7 satellite laser altimetry data, this paper studies the method of laser altimetry data processing, and preliminarily validates the geometric accuracy of laser altimetry standard products. Three regions are selected to validate the accuracy of GF-7 satellite laser altimetry standard products, which contain the domestic geometric calibration area, Huayin of Shaanxi province, North Rhine-Westphalia of Germany. The results show that the planimetric accuracy of laser altimetry standard products SLA03 in calibration area is (3.896±1.029)m and (3.286±0.337)m, and elevation accuracy is (0.018±0.099)m,(-0.017±0.096)m in respectively for the two-beam lasers. Moreover, the elevation control point quality control parameter ECP_Flag can effectively identify the laser points that can be used for elevation control. The overall accuracy of Shaanxi Huayin area is (-0.113±2.519)m and (0.191±1.071)m, respectively for the two beams. The elevation accuracy of laser points marked with ECP_Flag=1 is (0.111±0.152)m and (-0.064±0.115)m. The overall accuracy of Norwich in Germany is (-0.897±5.485)m and (-0.202 ±6.207)m, and the elevation accuracy of laser points marked with ECP_Flag=1 is (-0.304±0.190)m and (-0.279±0.220)m, respectively for the two beams. At present, the laser altimetry standard products of GF-7 satellite have been daily produced in the Land Satellite Application Center of the Ministry of Natural Resources.

In order to improve the accuracy of pixel-level change detection methods, this paper proposes a novel change detection method for remote sensing images by combining neighborhood information (including the neighborhood correlation image (NCI) and matching errors) and structural features. First, a technique of neighborhood correlation analysis is used to obtain the NCI which represents the context information, and the cross-correlation of neighborhood pixels is used to obtain matching errors by a template matching scheme. Then, structure features of images are extracted using orientated gradient information, which are robust to spectral differences between images. Subsequently, the initial change detection results is obtained by using the NCI, the matching errors, and structural features as the classification attributes of a decision tree. Finally, the Markov Random Field (MRF) is used to optimize the results, yielding the final binary map. The proposed method has been evaluated with two sets of bi-temporal remote sensing images from different sensors. Experimental results demonstrate that this method effectively improves the accuracy of change detection compared with the change vector analysis method, the single neighborhood information method and the method combining neighborhood information and texture features.

With the completion of the BeiDou-3 Navigation Satellite System (BDS-3) and the development of Galileo and quasi-zenith satellite system (QZSS), more and more satellites can be used to retrieve the atmospheric ionosphere. Generally, the short-time variation of receiver differential code biases (DCB) is considered as an important error source for ionospheric inversion using Global Navigation Satellite System (GNSS). However, some studies have shown that the short-time variation of receiver differential phase biases (DPB) may also affect the accuracy and reliability of ionospheric inversion. This paper presents a method to estimate the receiver DPB based on the single-differenced (SD) model without changing the reference satellite, which can realize the continuous estimation of the receiver DPB. DPB of the overlapping frequency combination of BDS-3, Galileo, GPS and QZSS is analyzed based on data collected from several multi-frequency and multi-constellation receivers capable of tracking the new signals of the BDS-3. The results show that ① The intraday changes of DPB of BDS-3, Galileo, GPS and QZSS are obvious and have a strong correlation with temperature. ② There is a strong correlation between the DPB of the overlapping frequency combinations of BDS-3, Galileo, GPS and QZSS. ③ There is a significant correlation between changes in DPB based on the baseline of the same type of combination.

Atmospheric precipitable water vapor (PWV) plays a key role in the study of global climate change and numerical weather prediction. However, the vertical correction of PWV mainly relies on a simple empirical correction model at present, which limite the high precision fusion of multi-source water vapor products and the comparisons between different water vapor products in a way. In this paper, for the characteristics of highly undulating terrain and diverse climate in China, the PWV vertical correction models which considering the time-varying lapse rate according to the whole areas in Mainland China and geographical divisions, named as C-PWVC1 model and C-PWVC2 model respectively, are established using ERA5 reanalysis data provided by the European Center for Medium Range Weather Forecast (ECMWF) from 2012 to 2017. Taking the profiles of 86 radiosonde stations in China in 2017 as reference values, the ERA5 and MERRA-2 reanalysis gridded data are interpolated into radiosonde stations to evaluate the performance of the newly established models, respectively. The results show that the performance of C-PWVC1 and C-PWVC2 models in PWV vertical correction is comparable. Compared with the condition without vertical correction, the performances of the C-PWVC model for ERA5 and MERRA-2 are improved by 16% and 8%, respectively. Compared with the commonly used PWV vertical correction model, C-PWVC model has little improvement for ERA5, but 12% for MERRA-2. C-PWVC model shows significant advantages when the height difference between two kinds of PWVs is large, and the correction capability for MERRA-2 is greater than ERA5. Moreover, for the ERA5 with different spatial resolutions, C-PWVC model has better interpolation precision and stability when compared with the commonly used PWV vertical correction model, especially in southern and western China. Therefore, the C-PWVC model has a good performance in PWV vertical correction for Mainland China, which can provide important application for the comparison and fusion of multi-source water vapor products in China.

Road markings are important traffic safety facilities. Its location, attribute, and topological relationship finely describe road traffic structure, and it is the basic data for applications such as intelligent traffic, high-precision maps, location, and navigation. This paper proposes a graph attention network with spatial context information (GAT_SCNet) to classify the road markings from mobile LiDAR point clouds. GAT_SCNet explores the graph structure to establish the appearance and dependence information among road markings. Meanwhile, GAT_SCNet incorporates the multi-head attention mechanism into the node propagation step, which computes the hidden states of each node based on the geometric, topological, and spatial structure relationships of the neighboring nodes. Finally, road markings classification is realized by the classification of nodes. Then, some schemes are designed for road markings vectorization. Four test datasets consisting of urban and highway scenes by different mobile laser scanning systems are used to evaluate the validities of the proposed method. Four accuracy evaluation metrics precision and recall of 9 types of road markings on the selected test datasets achieve (100.00%, 93.77%, 100.00%, 100.00%, 100.00%, 96.73%, 97.96%, 100.00%, 98.39%) and (100.00%, 96.36%, 100.00%, 10.000%, 100.00%, 97.26%, 85.72%, 100.00%, 94.16%), respectively. Accuracy evaluations and comparative studies prove that the proposed method has the capability of classifying multi-type road markings simultaneously and distinguishing similar road markings like dashed markings, zebra crossings, and stop lines in complex urban scenes.

Surveying and mapping production and services are facing many new problems such as real-time data acquisition, automation of information processing, and intellectualization of service applications. The main reason is that digital surveying and mapping involves complex algorithms and models as the core, and it is often impossible to use simple algorithms and models to completely describe and express the diverse, multi-dimensional and dynamic real world, which makes it difficult to address high-dimensional, nonlinear spatial problems. In order to address this challenge, it is necessary to explore the use of natural intelligence in surveying and mapping, and develop smart surveying and mapping technologies guided by knowledge and based on algorithms. This paper first discusses the basic concepts and ideas of smart surveying and mapping and then analyzes and points out the three basic problems of smart surveying and mapping, including the analysis and modeling of natural intelligence in surveying and mapping, the construction and realization of hybrid intelligent computing paradigm, and the mechanism and path of empowering production. The main directions for further development in the near future are then proposed, including the construction of knowledge system of smart surveying and mapping, research of technology methods, research and development of application systems and instrument equipment. In order to effectively promote scientific research and engineering applications in this area, efforts should be placed on top-level designing, promoting interdisciplinary collaborative innovation and industry-university-research cooperation, and creating good development environments.