Search

Ground-based multiple-input multiple-output synthetic aperture radar (MIMO-SAR) is a new type of deformation monitoring sensor that has the advantages of no mechanical motion and fast echo acquisition. Ground-based MIMO-SAR can significantly improve the data rate of deformation monitoring. In this paper, a fast imaging algorithm tailored for ground-based MIMO-SAR data is proposed, which can be applied in both far-field and near-field scenarios. First, the phase center approximation error of the non-collinear array in the near field is analyzed. Then, a fast imaging algorithm based on geometric transformation for the coherent synthesis of subimages is put forward. The algorithm uses the geometric transformation to convert the subaperture imaging results into the full aperture coordinate system, which avoids the point-by-point interpolation calculation and further reduces the computational cost of the subimage coherent synthesis algorithm. Simulations and experiments show that the algorithm can achieve high-precision focusing imaging, and its operation efficiency is significantly improved compared with the algorithm based on interpolation.

Construction safety accidents occur due to a combination of factors. Even a minor accident that could have been treated as a simple injury can lead to a serious accident or death, depending on when and where it occurred. Currently, methods for tracking worker behavior to manage such construction safety accidents are being studied. However, applying the methods to the construction site, various additional elements (e.g., sensors, transmitters, wearing equipment, and control systems) that must be additionally installed and managed are required. The cost of installation and management of these factors increases in proportion to the size of the site and the number of targets to be managed. In addition, the application of new equipment and new rules lowers the work efficiency of workers. In this paper, the following contents are described: (1) system overview, (2) image processing-QR code-based safety management target recognition methodology, and (3) object location discrimination technique applying the geometric transformation. Finally, the proposed methodology was tested to confirm the operation in the field, and the experimental results and conclusions were described in the paper.

Absolute localization of a flying UAV on its own in a global-navigation-satellite-system (GNSS)-denied environment is always a challenge. In this paper, we present a landmark-based approach where a UAV is automatically locked into the landmark scene shown in a georeferenced image via a feedback control loop, which is driven by the output of an aerial image registration. To pursue a real-time application, we design and implement a speeded-up-robust-features (SURF)-based image registration algorithm that focuses efficiency and robustness under a 2D geometric transformation. A linear UAV controller with signals of four degrees of freedom is derived from the estimated transformation matrix. The approach is validated in a virtual simulation environment, with experimental results demonstrating the effectiveness and robustness of the proposed UAV self-localization system.

This paper presents the development of a hybrid feature&mdash, dorsal hand vein and dorsal geometry&mdash, modality for human recognition. Our proposed hybrid feature extraction method exploits two types of features: dorsal hand geometric-related and local vein pattern. Using geometric affine invariants, the peg-free system extracts minutia points and vein termination and bifurcation and constructs a set of geometric invariants, which are then used to establish the correspondence between two sets of minutiae&mdash, one for the query vein image and the other for the reference vein image. When the correspondence is established, geometric transformation parameters are computed to align the query with the reference image. Once aligned, hybrid features are extracted for identification. In this study, the algorithm was tested on a database of 140 subjects, in which ten different dorsal hand geometric-related images were taken for each individual, and yielded the promising results. In this regard, we have achieved an equal error rate (EER) of 0.243%, indicating that our method is feasible and effective for dorsal vein recognition with high accuracy. This hierarchical scheme significantly improves the performance of personal verification and/or identification.

A visual&ndash, inertial odometer is used to fuse the image information obtained by a vision sensor with the data measured by an inertial sensor and recover the motion track online in a global frame. However, in an indoor environment, geometric transformation, sparse features, illumination changes, blurring, and noise will occur, which will either cause a reduction in or failure of the positioning accuracy. To solve this problem, a map matching algorithm based on an indoor plane structure map is proposed to improve the positioning accuracy of the system, this algorithm was implemented using a conditional random field model. The output of the attitude information from the visual&ndash, inertial odometer was used as the input of the conditional random field model. The feature function between the attitude information and the expected value was established, and the maximum probabilistic value of the attitude was estimated. Finally, the closed-loop feedback correction of the visual&ndash, inertial system was carried out with the probabilistic attitude value. A number of experiments were designed to verify the feasibility and reliability of the positioning method proposed in this paper.

The registration of point clouds in a three-dimensional space is an important task in many areas of computer vision, including robotics and autonomous driving. The purpose of registration is to find a rigid geometric transformation to align two point clouds. The registration problem can be affected by noise and partiality (two point clouds only have a partial overlap). The Iterative Closed Point (ICP) algorithm is a common method for solving the registration problem. Recently, artificial neural networks have begun to be used in the registration of point clouds. The drawback of ICP and other registration algorithms is the possible convergence to a local minimum. Thus, an important characteristic of a registration algorithm is the ability to avoid local minima. In this paper, we propose an ICP-type registration algorithm (λ-ICP) that uses a multiparameter functional (λ-functional). The proposed λ-ICP algorithm generalizes the NICP algorithm (normal ICP). The application of the λ-functional requires a consistent choice of the eigenvectors of the covariance matrix of two point clouds. The paper also proposes an algorithm for choosing the directions of eigenvectors. The performance of the proposed λ-ICP algorithm is compared with that of a standard point-to-point ICP and neural network Deep Closest Points (DCP).

Accurate recognition of tomato diseases is of great significance for agricultural production. Sufficient and insufficient training data of supervised recognition neural network training are symmetry problems. A high precision neural network needs a large number of labeled data, and the difficulty of data sample acquisition is the main challenge to improving the performance of disease recognition. The tomato leaf data augmented by the traditional data augmentation methods based on geometric transformation usually contain less information, and the generalization is not strong. Therefore, a new data augmentation method, RAHC_GAN, based on generative adversarial networks is proposed in this paper, which is used to expand tomato leaf data and identify diseases. In this method, continuous hidden variables are added at the input of the generator, and the purpose is to continuously control the size of the generated disease area and to supplement the intra class information of the same disease. Additionally, the residual attention block is added to the generator to make it pay more attention to the disease region in the leaf image; a multi-scale discriminator is also used to enrich the detailed texture of the generated image and finally generate leaves with obvious disease features. Then, we use the images generated by RAHC_GAN and the original training images to build an expanded data set, which is used to train four kinds of recognition networks, AlexNet, VGGNet, GoogLeNet, and ResNet, and the performance is evaluated through the test set. Experimental results show that RAHC_GAN can generate leaves with obvious disease features, and the generated expanded data set can significantly improve the recognition performance of the classifier. Furthermore, the results of the apple, grape, and corn data set show that RAHC_GAN can also be used as a method to solve the problem of insufficient data in other plant research tasks.

Most Facial Expression Recognition (FER) systems rely on machine learning approaches that require large databases for an effective training. As these are not easily available, a good solution is to augment the databases with appropriate data augmentation (DA) techniques, which are typically based on either geometric transformation or oversampling augmentations (e.g., generative adversarial networks (GANs)). However, it is not always easy to understand which DA technique may be more convenient for FER systems because most state-of-the-art experiments use different settings which makes the impact of DA techniques not comparable. To advance in this respect, in this paper, we evaluate and compare the impact of using well-established DA techniques on the emotion recognition accuracy of a FER system based on the well-known VGG16 convolutional neural network (CNN). In particular, we consider both geometric transformations and GAN to increase the amount of training images. We performed cross-database evaluations: training with the "augmented" KDEF database and testing with two different databases (CK+ and ExpW). The best results were obtained combining horizontal reflection, translation and GAN, bringing an accuracy increase of approximately 30%. This outperforms alternative approaches, except for the one technique that could however rely on a quite bigger database.

Accurate registration is an essential prerequisite for analysis and applications involving remote sensing imagery. It is usually difficult to extract enough matching points for inter-band registration in hyperspectral imagery due to the different spectral responses for land features in different image bands. This is especially true for non-adjacent bands. The inconsistency in geometric distortion caused by topographic relief also makes it inappropriate to use a single affine transformation relationship for the geometric transformation of the entire image. Currently, accurate registration between spectral bands of Zhuhai-1 satellite hyperspectral imagery remains challenging. In this paper, a full-spectrum registration method was proposed to address this problem. The method combines the transfer strategy based on the affine transformation relationship between adjacent spectrums with the differential correction from dense Delaunay triangulation. Firstly, the scale-invariant feature transform (SIFT) extraction method was used to extract and match feature points of adjacent bands. The RANdom SAmple Consensus (RANSAC) algorithm and the least square method is then used to eliminate mismatching point pairs to obtain fine matching point pairs. Secondly, a dense Delaunay triangulation was constructed based on fine matching point pairs. The affine transformation relation for non-adjacent bands was established for each triangle using the affine transformation relation transfer strategy. Finally, the affine transformation relation was used to perform differential correction for each triangle. Three Zhuhai-1 satellite hyperspectral images covering different terrains were used as experiment data. The evaluation results showed that the adjacent band registration accuracy ranged from 0.2 to 0.6 pixels. The structural similarity measure and cosine similarity measure between non-adjacent bands were both greater than 0.80. Moreover, the full-spectrum registration accuracy was less than 1 pixel. These registration results can meet the needs of Zhuhai-1 hyperspectral imagery applications in various fields.

This study builds robust hand shape features from the two modalities of depth and skeletal data for the dynamic hand gesture recognition problem. For the hand skeleton shape approach, we use the movement, the rotations of the hand joints with respect to their neighbors, and the skeletal point-cloud to learn the 3D geometric transformation. For the hand depth shape approach, we use the feature representation from the hand component segmentation model. Finally, we propose a multi-level feature LSTM with Conv1D, the Conv2D pyramid, and the LSTM block to deal with the diversity of hand features. Therefore, we propose a novel method by exploiting robust skeletal point-cloud features from skeletal data, as well as depth shape features from the hand component segmentation model in order for the multi-level feature LSTM model to benefit from both. Our proposed method achieves the best result on the Dynamic Hand Gesture Recognition (DHG) dataset with 14 and 28 classes for both depth and skeletal data with accuracies of 96.07% and 94.40%, respectively.

Mesh quality can affect both the accuracy and efficiency of numerical solutions. This paper first proposes a geometry-based smoothing and untangling method for 2D meshes based on explicit element geometric transformation and element stitching. A new explicit element geometric transformation (EEGT) operation for polygonal elements is firstly presented. The transformation, if applied iteratively to an arbitrary polygon (even inverted), will improve its regularity and quality. Then a well-designed element stitching scheme is introduced, which is achieved by carefully choosing appropriate element weights to average the temporary nodes obtained by the above individual element transformation. Based on the explicit element geometric transformation and element stitching, a new mesh smoothing and untangling approach for 2D meshes is proposed. The proper choice of averaging weights for element stitching ensures that the elements can be transitioned smoothly and uniformly throughout the calculation domain. Numerical results show that the proposed method is able to produce high-quality meshes with no inverted elements for highly tangled meshes. Besides, the inherent regularity and fine-grained parallelism make it suitable for implementation on Graphic Processor Unit (GPU).

Multispectral imagery contains abundant spectral information on terrestrial and oceanic targets, and retrieval of the geophysical variables of the targets is possible when the radiometric integrity of the data is secured. Multispectral cameras typically require the registration of individual band images because their lens locations for individual bands are often displaced from each other, thereby generating images of different viewing angles. Although this type of displacement can be corrected through a geometric transformation of the image coordinates, a mismatch or misregistration between the bands still remains, owing to the image acquisition timing that differs by bands. Even a short time difference is critical for the image quality of fast-moving targets, such as water surfaces, and this type of deformation cannot be compensated for with a geometric transformation between the bands. This study proposes a novel morphological band registration technique, based on the quantile matching method, for which the correspondence between the pixels of different bands is not sought by their geometric relationship, but by the radiometric distribution constructed in the vicinity of the pixel. In this study, a Micasense Rededge-M camera was operated on an unmanned aerial vehicle and multispectral images of coastal areas were acquired at various altitudes to examine the performance of the proposed method for different spatial scales. To assess the impact of the correction on a geophysical variable, the performance of the proposed method was evaluated for the chlorophyll-a concentration estimation. The results showed that the proposed method successfully removed the noisy spatial pattern caused by misregistration while maintaining the original spatial resolution for both homogeneous scenes and an episodic scene with a red tide outbreak.

The fusion of light detection and ranging (LiDAR) and camera data in real-time is known to be a crucial process in many applications, such as in autonomous driving, industrial automation, and robotics. Especially in the case of autonomous vehicles, the efficient fusion of data from these two types of sensors is important to enabling the depth of objects as well as the detection of objects at short and long distances. As both the sensors are capable of capturing the different attributes of the environment simultaneously, the integration of those attributes with an efficient fusion approach greatly benefits the reliable and consistent perception of the environment. This paper presents a method to estimate the distance (depth) between a self-driving car and other vehicles, objects, and signboards on its path using the accurate fusion approach. Based on the geometrical transformation and projection, low-level sensor fusion was performed between a camera and LiDAR using a 3D marker. Further, the fusion information is utilized to estimate the distance of objects detected by the RefineDet detector. Finally, the accuracy and performance of the sensor fusion and distance estimation approach were evaluated in terms of quantitative and qualitative analysis by considering real road and simulation environment scenarios. Thus the proposed low-level sensor fusion, based on the computational geometric transformation and projection for object distance estimation proves to be a promising solution for enabling reliable and consistent environment perception ability for autonomous vehicles.

Line of sight (LoS) and security are the fundamental requirement for any secure visual MIMO communication based vehicular networks. Uneven speed, irregular terrains, and uncertain appearance of obstacles disgrace LoS between the visual multiple-input multiple-output (MIMO) transmitter and receiver. This paper proposes optimal tracking techniques to maintain LoS using a novel quick response (QR) inspired light emitting array (LEA) pattern. Automatic geometric transformation of the QR code in conjunction with the particle filter is used to track the dynamically varying LEA position. In addition to that, an adaptive threshold technique (ATT) has been proposed to minimize error caused by ambient light variations and interferences. A novel key generation algorithm is also proposed to improve the security of the visual MIMO communication system. It utilizes a customized linear feedback shift register (LFSR) and synthetic color image to generate a stronger key. The randomness test demonstrates the robustness of the proposed key generation technique with a minimal probability value of 0.1223 (&gt, 0.01). Simulation and real-time studies have been carried out in the presence of geometric distortions, and the bit error rate (BER) performance is evaluated. Results illustrate that the proposed techniques make the visual MIMO system more secure and reliable for vehicular networks.

Phase correlation is one of the widely used image registration method in medical image processing and remote sensing. One of the main limitations of the phase correlation-based registration method is that it can only cope with Euclidean transformations, such as translation, rotation and scale, which constrain its application in wider fields, such as multi-view image matching, image-based navigation, etc. In this paper, we extended the phase correlation to perspective transformation by the combination of particle swarm optimization. Inspired by optic lens alignment based on interference, we propose to use the quality of PC fringes as the similarity, and then the aim of registration is to search for the optimized geometric transformation operator, which obtain the maximize value of PC-based similarity function through particle swarm optimization approach. The proposed method is validated by image registration experiments using simulated terrain shading, texture and natural landscape images containing different challenges, including illumination variation, lack of texture, motion blur, occlusion and geometric distortions. Further, image-based navigation experiments are carried out to demonstrate that the proposed method is able to correctly recover the trajectory of camera using multimodal target and reference image. Even under great radiometric and geometric distortions, the proposed method is able to achieve 0.1 sub-pixel matching accuracy on average while other methods fail to find the correspondence.

The tremendous advances in deep neural networks have demonstrated the superiority of deep learning techniques for applications such as object recognition or image classification. Nevertheless, deep learning-based methods usually require a large amount of training data, which mainly comes from manual annotation and is quite labor-intensive. In order to reduce the amount of manual work required for generating enough training data, we hereby propose to leverage existing labeled data to generate image annotations automatically. Specifically, the pixel labels are firstly transferred from one image modality to another image modality via geometric transformation to create initial image annotations, and then additional information (e.g., height measurements) is incorporated for Bayesian inference to update the labeling beliefs. Finally, the updated label assignments are optimized with a fully connected conditional random field (CRF), yielding refined labeling for all pixels in the image. The proposed approach is tested on two different scenarios, i.e., (1) label propagation from annotated aerial imagery to unmanned aerial vehicle (UAV) imagery and (2) label propagation from map database to aerial imagery. In each scenario, the refined image labels are used as pseudo-ground truth data for training a convolutional neural network (CNN). Results demonstrate that our model is able to produce accurate label assignments even around complex object boundaries, besides, the generated image labels can be effectively leveraged for training CNNs and achieve comparable classification accuracy as manual image annotations, more specifically, the per-class classification accuracy of the networks trained by the manual image annotations and the generated image labels have a difference within ± 5 % .

Co-registration is one of the most important steps in interferometric synthetic aperture radar (InSAR) data processing. The standard offset-measurement method based on cross-correlating uniformly distributed patches takes no account of specific geometric transformation between images or characteristics of ground scatterers. Hence, it is inefficient and difficult to obtain satisfying co-registration results for image pairs with relatively big distortion or large incoherent areas. Given this, an improved co-registration strategy is proposed in this paper which takes both the geometric features and image content into consideration. Firstly, some geometric transformations including scale, flip, rotation, and shear between images were eliminated based on the geometrical information, and the initial co-registration polynomial was obtained. Then the registration points were automatically detected by integrating the signal-to-clutter-ratio (SCR) thresholds and the amplitude information, and a further co-registration process was performed to refine the polynomial. Several comparison experiments were carried out using 2 TerraSAR-X data from the Hong Kong airport and 21 PALSAR data from the Donghai Bridge. Experiment results demonstrate that the proposed method brings accuracy and efficiency improvements for co-registration and processing abilities in the cases of big distortion between images or large incoherent areas in the images. For most co-registrations, the proposed method can enhance the reliability and applicability of co-registration and thus promote the automation to a higher level.

With the rapid improvement of geospatial data acquisition and processing techniques, a variety of geospatial databases from public or private organizations have become available. Quite often, one dataset may be superior to other datasets in one, but not all aspects. In Germany, for instance, there were three major road network vector data, viz. Tele Atlas (which is now “TOMTOM”), NAVTEQ (which is now “here”), and ATKIS. However, none of them was qualified for the purpose of multi-modal navigation (e.g., driving + walking): Tele Atlas and NAVTEQ consist of comprehensive routing-relevant information, but many pedestrian ways are missing; ATKIS covers more pedestrian areas but the road objects are not fully attributed. To satisfy the requirements of multi-modal navigation, an automatic approach has been proposed to conflate different road networks together, which involves five routines: (a) road-network matching between datasets; (b) identification of the pedestrian ways; (c) geometric transformation to eliminate geometric inconsistency; (d) topologic remodeling of the conflated road network; and (e) error checking and correction. The proposed approach demonstrates high performance in a number of large test areas and therefore has been successfully utilized for the real-world data production in the whole region of Germany. As a result, the conflated road network allows the multi-modal navigation of “driving + walking”.

The integration of underwater 3D data captured by acoustic and optical systems is a promising technique in various applications such as mapping or vehicle navigation. It allows for compensating the drawbacks of the low resolution of acoustic sensors and the limitations of optical sensors in bad visibility conditions. Aligning these data is a challenging problem, as it is hard to make a point-to-point correspondence. This paper presents a multi-sensor registration for the automatic integration of 3D data acquired from a stereovision system and a 3D acoustic camera in close-range acquisition. An appropriate rig has been used in the laboratory tests to determine the relative position between the two sensor frames. The experimental results show that our alignment approach, based on the acquisition of a rig in several poses, can be adopted to estimate the rigid transformation between the two heterogeneous sensors. A first estimation of the unknown geometric transformation is obtained by a registration of the two 3D point clouds, but it ends up to be strongly affected by noise and data dispersion. A robust and optimal estimation is obtained by a statistical processing of the transformations computed for each pose. The effectiveness of the method has been demonstrated in this first experimentation of the proposed 3D opto-acoustic camera.

Coordinate transformation plays an indispensable role in industrial measurements, including photogrammetry, geodesy, laser 3-D measurement and robotics. The widely applied methods of coordinate transformation are generally based on solving the equations of point clouds. Despite the high accuracy, this might result in no solution due to the use of ill conditioned matrices. In this paper, a novel coordinate transformation method is proposed, not based on the equation solution but based on the geometric transformation. We construct characteristic lines to represent the coordinate systems. According to the space geometry relation, the characteristic line scan is made to coincide by a series of rotations and translations. The transformation matrix can be obtained using matrix transformation theory. Experiments are designed to compare the proposed method with other methods. The results show that the proposed method has the same high accuracy, but the operation is more convenient and flexible. A multi-sensor combined measurement system is also presented to improve the position accuracy of a robot with the calibration of the robot kinematic parameters. Experimental verification shows that the position accuracy of robot manipulator is improved by 45.8% with the proposed method and robot calibration.

Displacement and deformation are fundamental measures of Earth surface mass movements such as glacier flow, rockglacier creep and rockslides. Ground-based methods of monitoring such mass movements can be costly, time consuming and limited in spatial and temporal coverage. Remote sensing techniques, here matching of repeat optical images, are increasingly used to obtain displacement and deformation fields. Strain rates are usually computed in a post-processing step based on the gradients of the measured velocity field. This study explores the potential of automatically and directly computing velocity, rotation and strain rates on Earth surface mass movements simultaneously from the matching positions and the parameters of the geometric transformation models using the least squares matching (LSM) approach. The procedures are exemplified using bi-temporal high resolution satellite and aerial images of glacier flow, rockglacier creep and land sliding. The results show that LSM matches the images and computes longitudinal strain rates, transverse strain rates and shear strain rates reliably with mean absolute deviations in the order of 10−4 (one level of significance below the measured values) as evaluated on stable grounds. The LSM also improves the accuracy of displacement estimation of the pixel-precision normalized cross-correlation by over 90% under ideal (simulated) circumstances and by about 25% for real multi-temporal images of mass movements.

Made available in DSpace on 2019-10-06T16:09:02Z (GMT). No. of bitstreams: 0 Previous issue date: 2018-12-01 Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) A technique for the reconstruction of cylindrical surfaces using optical images with an extension of least squares matching is presented. This technique is based on stereo-image acquisition of a cylindrical object, and it involves displacing the camera following the object length. The basic concept behind this technique is that variations in the camera viewpoint over a cylindrical object produce perspective effects similar to a conic section in an image sequence. Such parallax changes are continuous and can be modelled by a second-order function, which is combined with an adaptive least squares matching (ALSM) for the 3D object reconstruction. Using this concept, a photogrammetric intersection with only two image patches can be used to model a cylindrical object with high accuracy. Experiments were conducted with a cylinder on a panel with coded targets to assess the 3D reconstruction accuracy. The accuracy assessment was based on a comparison between the estimated diameter and the diameter directly measured over the cylinder. The difference between the diameters indicated an accuracy of 1/10 mm, and the cylindrical surface was entirely reconstructed. Department of Statistics São Paulo State University UNESP, 305 Department of Cartography São Paulo State University UNESP, 305 Department of Statistics São Paulo State University UNESP, 305 Department of Cartography São Paulo State University UNESP, 305 FAPESP: 2013/50426-4 FAPESP: 2014/05033-7

Nowadays, three-dimensional (3D) printing technology is applied to many areas of life and changes the world based on the creation of complex structures and shapes that were not feasible in the past. But, the data of 3D printing is often attacked in the storage and transmission processes. Therefore, 3D printing must be ensured security in the manufacturing process, especially the data of 3D printing to prevent attacks from hackers. This paper presents a security solution for 3D printing based on two-dimensional (2D) slices encryption. The 2D slices of 3D printing data is encrypted in the frequency domain or in the spatial domain by the secret key to generate the encrypted data of 3D printing. We implemented the proposed solution in both the frequency domain based on the Discrete Cosine Transform and the spatial domain based on geometric transform. The entire 2D slices of 3D printing data is altered and secured after the encryption process. The proposed solution is responsive to the security requirements for the secured storage and transmission. Experimental results also verified that the proposed solution is effective to 3D printing data and is independent on the format of 3D printing models. When compared to the conventional works, the security and performance of the proposed solution is also better.