Search

Copyright of Journal of South China Agricultural University is the property of Gai Kan Bian Wei Hui and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Tiangong University is the property of Journal of Tianjin Polytechnic University and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

To help Latin dancers feel the rhythm of the sports dance, it is necessary to standardize the posture action during basic training. Therefore, it is important to study the method of correcting Latin dance posture action. The traditional Latin dance posture correction methods have some problems, such as the included angle error of head motion, the angle error of spine transformation, the error of fit between foot and ground and so on. In this paper, a Latin dance posture correction method using improved deep reinforcement learning in the Internet of things (IoT) is proposed. First, the Latin dance posture image acquisition architecture is constructed using IoT and binocular stereo vision to acquire Latin dance posture images and extract Latin dance posture features. Second, the channel attention module in the deep learning network is improved, and the Latin dance posture diagnosis model is constructed based on the action feature extraction results using the improved deep robust chemical network. Finally, the action correction coefficients are calculated according to the Latin dance posture diagnosis results to realize the Latin dance posture correction. The results showed that after the application of the proposed correction method, including angle error of head movement, the spine transformation angle error and fit between foot and ground error of the participants’ motions were kept below 1∘, and the frame position offset was 1.3cm. It indicates that the proposed method can effectively improve the degree of Latin dance posture specification. [ABSTRACT FROM AUTHOR]

Accurate three-dimensional (3D) measurement for large field of view (FOV) is currently a significant research field. Accordingly, system calibration is crucial to ensure accuracy. However typical calibration methods often involve the use of large calibration objects, which is not only expensive but also difficult to achieve sufficient accuracy. A novel method based on a dual-brand deep neural network (DNN) is proposed for the system calibration. Taking advantage of the concept of "divide and conquer", the FOV is divided into sub-regions with a part of overlapping regions by a small calibration object, which forms a large calibration object covering the whole FOV. Then the sub-regions are fused into a global framework and further optimized by the proposed dual-brand DNN. The proposed method reduces the need for calibration objects while improving the calibration accuracy and generalization ability in large FOV. A series of experiments have been designed to prove the effectiveness and robustness of the proposed method. • A high-accuracy calibration method of a 3D measurement system with large field-of-view is proposed. The highlights of the paper are: • A high-accuracy calibration method for large field-of-view is proposed. • The large field-of-view is divided into multiple sub-regions. • Sub-regions are unified to a uniform coordinate system by self-constraint. • A dual-branch DNN is optimized using the multi-objective loss function. [ABSTRACT FROM AUTHOR]

The non-destructive measurement of fish is an important link in intelligent aquaculture, and realizing the accurate estimation of fish mass is the key to the stable operation of this link. Taking tilapia as the object, this study proposes an underwater tilapia mass estimation method, which can accurately estimate the mass of free-swimming tilapia under non-contact conditions. First, image enhancement is performed on the original image, and the depth image is obtained by correcting and stereo matching the enhanced image using binocular stereo vision technology. And the fish body is segmented by an SAM model. Then, the segmented fish body is labeled with key points, thus realizing the 3D reconstruction of tilapia. Five mass estimation models are established based on the relationship between the body length and the mass of tilapia, so as to realize the mass estimation of tilapia. The results showed that the average relative errors of the method models were 5.34%~7.25%. The coefficient of determination of the final tilapia mass estimation with manual measurement was 0.99, and the average relative error was 5.90%. The improvement over existing deep learning methods is about 1.54%. This study will provide key technical support for the non-destructive measurement of tilapia, which is of great significance to the information management of aquaculture, the assessment of fish growth condition, and baiting control. [ABSTRACT FROM AUTHOR]

Large hoop truss antennas are locked into a stable structure during deployment in space. The key to characterizing the performance of each mechanism, and discovering problems in the deployment process, is the three-dimensional motion of the antenna. Therefore, it is important to measure antenna motion with high precision for antenna design. Contact measurement requires sensors to be attached to the object, affecting the motion of the antenna; thus, noncontact measurement is the preferred measurement method. For a large antenna, a large field of view and observation angle are required to measure the deployment process. Traditional binocular stereo vision measurement is not sufficiently accurate. In this paper, 36 cameras were combined into an array to measure antenna deployment. A high-precision triggering system was designed, a calibration method suitable for large-scale, multiangle measurement was developed, and a reconstruction method based on the geometric constraints was proposed, greatly improving measurement resolution. The camera array was used to measure the motion of a large hoop truss antenna with a diameter of 4 m during deployment. Based on the measured parameters, the angular velocity of each antenna mechanism was calculated, and the synchronization of the antenna structure was analyzed. [ABSTRACT FROM AUTHOR]

Due to the operation of live-line working robots at elevated heights for precision tasks, a suitable visual assistance system is essential to determine the position and distance of the robotic arm or gripper relative to the target object. In this study, we propose a method for distance measurement in live-line working robots by integrating the YOLOv5 algorithm with binocular stereo vision. The camera's intrinsic and extrinsic parameters, as well as distortion coefficients, are obtained using the Zhang Zhengyou calibration method. Subsequently, stereo rectification is performed on the images to establish a standardized binocular stereovision model. The Census and Sum of Absolute Differences (SAD) fused stereo matching algorithm is applied to compute the disparity map. We train a dataset of transmission line bolts within the YOLO framework to derive the optimal model. The identified bolts are framed, and the depth distance of the target is ultimately calculated. And through the experimental verification of the bolt positioning, the results show that the method can achieve a relative error of 1% in the proximity of positioning. This approach provides real-time and accurate environmental perception for symmetrical structural live-line working robots, enhancing the stability of these robots. [ABSTRACT FROM AUTHOR]

The advancement of unloading technology in combine harvesting is crucial for the intelligent development of agricultural machinery. Accurately measuring material pile height in transport vehicles is essential, as uneven accumulation can lead to spillage and voids, reducing loading efficiency. Relying solely on manual observation for measuring stack height can decrease harvesting efficiency and pose safety risks due to driver distraction. This research applies binocular vision to agricultural harvesting, proposing a novel method that uses a stereo matching algorithm to measure material pile height during harvesting. By comparing distance measurements taken in both empty and loaded states, the method determines stack height. A linear regression model processes the stack height data, enhancing measurement accuracy. A binocular vision system was established, applying Zhang’s calibration method on the MATLAB (R2019a) platform to correct camera parameters, achieving a calibration error of 0.15 pixels. The study implemented block matching (BM) and semi-global block matching (SGBM) algorithms using the OpenCV (4.8.1) library on the PyCharm (2020.3.5) platform for stereo matching, generating disparity, and pseudo-color maps. Three-dimensional coordinates of key points on the piled material were calculated to measure distances from the vehicle container bottom and material surface to the binocular camera, allowing for the calculation of material pile height. Furthermore, a linear regression model was applied to correct the data, enhancing the accuracy of the measured pile height. The results indicate that by employing binocular stereo vision and stereo matching algorithms, followed by linear regression, this method can accurately calculate material pile height. The average relative error for the BM algorithm was 3.70%, and for the SGBM algorithm, it was 3.35%, both within the acceptable precision range. While the SGBM algorithm was, on average, 46 ms slower than the BM algorithm, both maintained errors under 7% and computation times under 100 ms, meeting the real-time measurement requirements for combine harvesting. In practical operations, this method can effectively measure material pile height in transport vehicles. The choice of matching algorithm should consider container size, material properties, and the balance between measurement time, accuracy, and disparity map completeness. This approach aids in manual adjustment of machinery posture and provides data support for future autonomous master-slave collaborative operations in combine harvesting.

In the process of grading and dynamically optimizing the design and construction parameters of the surrounding rock mass of a rock tunnel face, efficiently and accurately acquiring the geometrical parameters of the rock discontinuities is an important basic task. To address the problems of time consuming, low accuracy, and high danger associated with traditional methods of obtaining the structural information of rock mass, this paper proposes a method for three-dimensional reconstruction and intelligent information extraction of tunnel face based on binocular stereo vision (BSV). First, the parallel binocular device with a single camera was improved, calibrated using the checkerboard calibration method. By integrating with the semi-global matching algorithm, the BSV based method for the three-dimensional reconstruction of the rock mass of the tunnel face was optimized. Furthermore, based on the results from on-site engineering applications, this study leveraged two parameters, point cloud density and algorithm runtime, to determine the optimal values for the disparity range and window size parameters within the semi-global stereo matching algorithm. This enhancement improved the performance of the 3D reconstruction method based on binocular stereo vision. Finally, efficient and refined intelligent methods for extracting structural parameters of the rock mass were proposed based on k-nearest neighbor search and kernel density estimation. The research results can provide reliable technical support for the intelligent and efficient acquisition of rock mass structural information in rock tunnel engineering faces. [ABSTRACT FROM AUTHOR]

Copyright of Electric Drive for Locomotives is the property of Electric Drive for Locomotives Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The deformation detection of large machinery is usually achieved using threedimensional displacement measurement. Binocular stereo vision measurement technology, as a commonly used digital image correlation method, has received widespread attention in the academic community. Binocular stereo vision achieves the goal of three-dimensional displacement measurement by simulating the working mode of the human eyes, but the measurement is easily affected by light refraction. Based on this, the study introduces particle swarm optimization algorithm for target displacement measurement on Canon imaging dataset, and introduces backpropagation neural network for mutation processing of particles in particle swarm algorithm to generate fusion algorithm. It combines the four coordinate systems of world, pixel, physics, and camera to establish connections. Taking into account environmental factors and lens errors, the camera parameters and deformation coefficients were revised by shooting a black and white checkerboard. Finally, the study first conducted error analysis on binocular stereo vision technology in three dimensions, and the relative error remained stable at 1 % within about 60 seconds. At the same time, three algorithms, including the spotted hyena algorithm, were introduced to conduct performance comparison experiments using particle swarm optimization and backpropagation network algorithms. The experiment shows that the three-dimensional error of the fusion algorithm gradually stabilizes within the range of [-0.5 %, 0.5 %] over time, while the two-dimensional error generally hovers around 0 value. Its performance is significantly superior to other algorithms, so the binocular stereo vision of this fusion algorithm can achieve good measurement results. [ABSTRACT FROM AUTHOR]

(1) Background: In real road scenarios, various complex environmental conditions may occur, including bright lights, nighttime, rain, and snow. In such a complex environment for detecting pole-shaped obstacles, it is easy to lose the feature information. A high rate of leakage detection, false positives, and measurement errors are generated as a result. (2) Methods: The first part of this paper utilizes the improved YOLOv5 algorithm to detect and classify pole-shaped obstacles. Then, the identified target frame information is combined with binocular stereo matching to obtain more accurate distance information. (3) Results: The experimental results demonstrate that this method achieves a mean average precision (mAP) of 97.4% for detecting pole-shaped obstacles, which is 3.1% higher than the original model. The image inference time is only 1.6 ms, which is 1.8 ms faster than the original algorithm. Additionally, the model size is only 19.0 MB. Furthermore, the range error of this system is less than 7% within the range of 3–15 m. (4) Conclusions: Therefore, the algorithm not only achieves real-time and accurate identification and classification but also ensures precise measurement within a specific range. Meanwhile, the model is lightweight and better suited for deploying sensing systems. [ABSTRACT FROM AUTHOR]

Abstract Fast and accurate measurement of the volume of earthmoving materials is of great significance for the real-time evaluation of loader operation efficiency and the realization of autonomous operation. Existing methods for volume measurement, such as total station-based methods, cannot measure the volume in real time, while the bucket-based method also has the disadvantage of poor universality. In this study, a fast estimation method for a loader’s shovel load volume by 3D reconstruction of material piles is proposed. First, a dense stereo matching method (QORB–MAPM) was proposed by integrating the improved quadtree ORB algorithm (QORB) and the maximum a posteriori probability model (MAPM), which achieves fast matching of feature points and dense 3D reconstruction of material piles. Second, the 3D point cloud model of the material piles before and after shoveling was registered and segmented to obtain the 3D point cloud model of the shoveling area, and the Alpha-shape algorithm of Delaunay triangulation was used to estimate the volume of the 3D point cloud model. Finally, a shovel loading volume measurement experiment was conducted under loose-soil working conditions. The results show that the shovel loading volume estimation method (QORB–MAPM VE) proposed in this study has higher estimation accuracy and less calculation time in volume estimation and bucket fill factor estimation, and it has significant theoretical research and engineering application value.

Abstract The binocular stereo vision is the lowest cost sensor for obtaining 3D information. Considering the weakness of long‐distance measurement and stability, the improvement of accuracy and stability of stereo vision is urgently required for application of precision agriculture. To address the challenges of stereo vision long‐distance measurement and stable perception without hardware upgrade, inspired by hawk eyes, higher resolution perception and the adaptive HDR (High Dynamic Range) were introduced in this paper. Simulating the function from physiological structure of ‘deep fovea’ and ‘shallow fovea’ of hawk eye, the higher resolution reconstruction method in this paper was aimed at accuracy improving. Inspired by adjustment of pupils, the adaptive HDR method was proposed for high dynamic range optimisation and stable perception. In various light conditions, compared with default stereo vision, the accuracy of proposed algorithm was improved by 28.0% evaluated by error ratio, and the stability was improved by 26.56% by disparity accuracy. For fixed distance measurement, the maximum improvement was 78.6% by standard deviation. Based on the hawk‐eye‐inspired perception algorithm, the point cloud of orchard was improved both in quality and quantity. The hawk‐eye‐inspired perception algorithm contributed great advance in binocular 3D point cloud reconstruction in orchard navigation map.

Under the impact of the wave of new technology, the form of piano teaching is no longer limited to the classroom and paper textbooks, and modern information technology, such as the Internet and virtual reality, is constantly integrated into traditional piano teaching. This paper utilizes the Unity3D engine to create a virtual reality environment for piano teaching interaction, based on VR technology, and to construct an interactive platform for piano teaching. Firstly, based on binocular stereo vision for the three-dimensional reconstruction of the teaching interactive scene, based on the improvement of the LeNet5 network for the user’s head gesture tracking, the constructed virtual scene realizes the real-time movement with the movement of the human head. The test set images using CPU-i76700HQ and GTX960 graphics cards consume 78ms and 25ms per frame, respectively, which meets the real-time system requirements. In this paper, the improved LeNet5 algorithm for head pose estimation without face detection is applicable and has a high detection accuracy of 99.31%, which is better than the algorithms such as Adaboost+Hough, LBF+BP, CNN+Heatmaps, DCNN, and FSA-Net, and is able to satisfy the real-time detection requirements of the virtual scene. The virtual piano teaching interactive platform has an average rating of more than 4 in attractiveness, subjectivity, interactivity, and convenience, but the running speed is only 3.46, so it is necessary to strengthen the technical support and improve the software and hardware level to break through the development bottleneck. Traditional piano teaching is given new vitality and kinetic energy by the interactive platform constructed in this paper for piano teaching in a virtual reality environment.

Art and design professional education can not keep up with the development speed of the industry, and there is no reference to the market demand data, failing to keep pace with the times, which brings many negative impacts on the employment of students. This paper suggests an art design professional education model that incorporates VR interactive scene technology for this reason. Through the double camera, we simulate the visual principle of the human eye, collect three-dimensional data from VR interactive scenes, and use the image coordinate system and pixel coordinate system conversion operations to pre-process the collected data. The optimal layout solution is sought by adjusting the layout to generate a realistic 3D model of the teaching scene and to realize the construction of the VR interactive scene. The principle and implementation process of the art professional education model supported by the VR interactive scene is elaborated, and the experimental comparison method is used to empirically analyze the art and design professional education model integrating VR technology. The results show that there is a significant difference between the experimental group and the control group in the dimensions of adaptability (0.004) and uniqueness (0.044) of design thinking and the cultivation of design thinking ability (0.016) (P

As an important part of university physical education, table tennis teaching occupies a crucial position in physical education reform and requires continuous innovation and reform. This paper first starts from the current situation of table tennis teaching and proposes the application of a binocular stereo vision system based on a binocular stereo vision system for university table tennis teaching, which lays the foundation for three-dimensional reconstruction and skill learning. Secondly, the mathematical model of the 3D reconstruction algorithm is constructed by the tangential distortion of least squares, which avoids the complex algorithm of iteratively solving the 3D world coordinates of table tennis and improves the speed of the 3D reconstruction algorithm while the accuracy meets the subsequent prediction requirements. Finally, students in the experimental and control groups were compared in the technical evaluation of table tennis backhand pushing, attacking techniques, and left pushing and right attacking to verify the effectiveness of the table tennis binocular stereo vision teaching system. The study results showed that the experimental group students trained by the binocular stereo vision system for table tennis had 41.77% higher accuracy, 23.13% higher coordination, 12.31% higher return speed, and 14.84% higher hitting power compared to the control group. Applying a binocular stereo-vision system to table tennis teaching is novel and is conducive to stimulating students’ curiosity and challenge, enhancing their interest in learning, and effectively improving their understanding and knowledge of table tennis techniques.

Fast and accurate measurement of the volume of earthmoving materials is of great significance for the real-time evaluation of loader operation efficiency and the realization of autonomous operation. Existing methods for volume measurement, such as total station-based methods, cannot measure the volume in real time, while the bucket-based method also has the disadvantage of poor universality. In this study, a fast estimation method for a loader's shovel load volume by 3D reconstruction of material piles is proposed. First, a dense stereo matching method (QORB–MAPM) was proposed by integrating the improved quadtree ORB algorithm (QORB) and the maximum a posteriori probability model (MAPM), which achieves fast matching of feature points and dense 3D reconstruction of material piles. Second, the 3D point cloud model of the material piles before and after shoveling was registered and segmented to obtain the 3D point cloud model of the shoveling area, and the Alpha-shape algorithm of Delaunay triangulation was used to estimate the volume of the 3D point cloud model. Finally, a shovel loading volume measurement experiment was conducted under loose-soil working conditions. The results show that the shovel loading volume estimation method (QORB–MAPM VE) proposed in this study has higher estimation accuracy and less calculation time in volume estimation and bucket fill factor estimation, and it has significant theoretical research and engineering application value. [ABSTRACT FROM AUTHOR]

Three-dimensional reconstruction technology based on binocular stereo vision is a key research area with potential clinical applications. Mainstream research has focused on sparse point reconstruction within the soft tissue domain, limiting the comprehensive 3D data acquisition required for effective surgical robot navigation. This study introduces a new paradigm to address existing challenges. An innovative stereoscopic endoscopic image correction algorithm is proposed, exploiting intrinsic insights into stereoscopic calibration parameters. The synergy between the stereoscopic endoscope parameters and the disparity map derived from the cardiac soft tissue images ultimately leads to the acquisition of precise 3D points. Guided by deliberate filtering and optimization methods, the triangulation process subsequently facilitates the reconstruction of the complex surface of the cardiac soft tissue. The experimental results strongly emphasize the accuracy of the calibration algorithm, confirming its utility in stereoscopic endoscopy. Furthermore, the image rectification algorithm exhibits a significant reduction in vertical parallax, which effectively enhances the stereo matching process. The resulting 3D reconstruction technique enables the targeted surface reconstruction of different regions of interest in the cardiac soft tissue landscape. This study demonstrates the potential of binocular stereo vision-based 3D reconstruction techniques for integration into clinical settings. The combination of joint calibration algorithms, image correction innovations, and precise tissue reconstruction enhances the promise of improved surgical precision and outcomes in the field of cardiac interventions. [ABSTRACT FROM AUTHOR]

The binocular stereo vision is the lowest cost sensor for obtaining 3D information. Considering the weakness of long‐distance measurement and stability, the improvement of accuracy and stability of stereo vision is urgently required for application of precision agriculture. To address the challenges of stereo vision long‐distance measurement and stable perception without hardware upgrade, inspired by hawk eyes, higher resolution perception and the adaptive HDR (High Dynamic Range) were introduced in this paper. Simulating the function from physiological structure of 'deep fovea' and 'shallow fovea' of hawk eye, the higher resolution reconstruction method in this paper was aimed at accuracy improving. Inspired by adjustment of pupils, the adaptive HDR method was proposed for high dynamic range optimisation and stable perception. In various light conditions, compared with default stereo vision, the accuracy of proposed algorithm was improved by 28.0% evaluated by error ratio, and the stability was improved by 26.56% by disparity accuracy. For fixed distance measurement, the maximum improvement was 78.6% by standard deviation. Based on the hawk‐eye‐inspired perception algorithm, the point cloud of orchard was improved both in quality and quantity. The hawk‐eye‐inspired perception algorithm contributed great advance in binocular 3D point cloud reconstruction in orchard navigation map. [ABSTRACT FROM AUTHOR]

The non-destructive measurement of fish is an important link in intelligent aquaculture, and realizing the accurate estimation of fish mass is the key to the stable operation of this link. Taking tilapia as the object, this study proposes an underwater tilapia mass estimation method, which can accurately estimate the mass of free-swimming tilapia under non-contact conditions. First, image enhancement is performed on the original image, and the depth image is obtained by correcting and stereo matching the enhanced image using binocular stereo vision technology. And the fish body is segmented by an SAM model. Then, the segmented fish body is labeled with key points, thus realizing the 3D reconstruction of tilapia. Five mass estimation models are established based on the relationship between the body length and the mass of tilapia, so as to realize the mass estimation of tilapia. The results showed that the average relative errors of the method models were 5.34%~7.25%. The coefficient of determination of the final tilapia mass estimation with manual measurement was 0.99, and the average relative error was 5.90%. The improvement over existing deep learning methods is about 1.54%. This study will provide key technical support for the non-destructive measurement of tilapia, which is of great significance to the information management of aquaculture, the assessment of fish growth condition, and baiting control.

Due to the operation of live-line working robots at elevated heights for precision tasks, a suitable visual assistance system is essential to determine the position and distance of the robotic arm or gripper relative to the target object. In this study, we propose a method for distance measurement in live-line working robots by integrating the YOLOv5 algorithm with binocular stereo vision. The camera’s intrinsic and extrinsic parameters, as well as distortion coefficients, are obtained using the Zhang Zhengyou calibration method. Subsequently, stereo rectification is performed on the images to establish a standardized binocular stereovision model. The Census and Sum of Absolute Differences (SAD) fused stereo matching algorithm is applied to compute the disparity map. We train a dataset of transmission line bolts within the YOLO framework to derive the optimal model. The identified bolts are framed, and the depth distance of the target is ultimately calculated. And through the experimental verification of the bolt positioning, the results show that the method can achieve a relative error of 1% in the proximity of positioning. This approach provides real-time and accurate environmental perception for symmetrical structural live-line working robots, enhancing the stability of these robots.

To achieve the rapid recognition and accurate picking of Camellia oleifera fruits, a binocular vision system composed of two industrial cameras was used to collect images of Camellia oleifera fruits in natural environments. The YOLOv7 convolutional neural network model was used for iterative training, and the optimal weight model was selected to recognize the images and obtain the anchor frame region of the Camellia oleifera fruits. The local binary pattern (LBP) maps of the anchor frame region were extracted and matched by using the normalized correlation coefficient template matching algorithm to obtain the positions of the center point in the left and right images. The recognition experimental results showed that the accuracy rate, recall rate, mAP and F1 of the model were 97.3%, 97.6%, 97.7% and 97.4%. The recognition rate of the Camellia oleifera fruit with slight shading was 93.13%, and the recognition rate with severe shading was 75.21%. The recognition rate of the Camellia oleifera fruit was 90.64% under sunlight condition, and the recognition rate was 91.34% under shading condition. The orchard experiment results showed that, in the depth range of 400–600 mm, the maximum error value of the binocular stereo vision system in the depth direction was 4.279 mm, and the standard deviation was 1.142 mm. The detection and three-dimensional positioning accuracy of the binocular stereo vision system for Camellia oleifera fruits could basically meet the working requirements of the Camellia oleifera fruit-picking robot. [ABSTRACT FROM AUTHOR]

To solve the problem of online detection of coal-carrying rate in gangue, an approach based on particle queuing and binocular machine vision was proposed. First, a queuing device was applied to line up the coal and gangue particles. A binocular camera was used to obtain images of coal and gangue from above the belt. Through image segmentation, the coal and gangue particle regions were divided from the belt background. Then, the Semi-Global Block Matching (SGBM) algorithm was used to obtain the height information matrix of each region. At the same time, the projected area of the particles was calculated from the area of the divided regions. It was converted from pixel units to physical units by capturing an image of a ruler. The height matrix and projected area were combined to construct a mechanism-based volume model by integral principle. Finally, the calculated volume of coal and gangue were multiplied by their respective empirical density to get the mass of particles. The coal-carrying rate in gangue was obtained by dividing the mass of coal by the total mass. The average relative error of the proposed method is 2.79%. It not only overcomes the prediction error caused by particle stacking, but also improves the computational efficiency of image processing, and the "white-box model" is also easy to understand. [ABSTRACT FROM AUTHOR]

The core aim of motion-recognition technology is to detect moving targets and extract the target's movement characteristics, classifying these so as to judge the corresponding action category. Moving target detection and tracking form the core of computer vision research topics in military visual guidance, robot visual navigation, industrial product detection, medical diagnosis, traffic monitoring, and virtual reality, and is also very important in terms of practical value and the broad prospects for development in other fields. However, in human behavior recognition algorithms, the background, light, environment, and other factors make identification accuracy difficult to obtain and unstable, with limited application scenarios. Because traditional target detection and tracking is based on monocular vision algorithm robustness, which is poor, this makes it difficult to detect and track moving targets in complex environments. In recent years, with the maturity of target detection and visual range-finding technology, pure static target visual barium distance or the depth of target detection cannot meet the demand of the market. Therefore, we need the efficient collection of deep information in moving target detection systems. Binocular stereo vision technology, as a hot research field in computer vision, can be widely used in the field of sports to achieve target detection and ranging functions. In view of this, this paper adopted a stereoscopic vision for use in target detection and tracking methods. This paper proposes a stereo-matching method based on hierarchical network minimum segmentation, where the method uses three-dimensional images from two layers of the pyramid decomposition by splitting the global optimal search strategy into a minimum at the top of the image to match with the initial parallax at the bottom of the image through regional gray correlation matching. Both reduce the search space of the stereo matching and improve matching accuracy. At the same time, this paper uses the motion of a football shot to prospect this recognition method, which can identify the target area. The Snake contour model method was also utilized to extract the target contour. This method has the following three characteristics: when ambient light changes slowly and suddenly, it can correctly detect moving targets; when the object is lit with cast shadows, the moving target can be detected correctly and is not affected by the shadows. When rigid target motion and deformation occur, it can correctly extract the target contour. [ABSTRACT FROM AUTHOR]

At present, underwater exploration and salvage, underwater archaeology, and other underwater operations still mainly rely on professional underwater operators. Considering that artificial underwater operation is faced with the problems of small exploration scope, poor working environment, and low work efficiency, it is the future trend to use robots to replace manual underwater operation in related fields. Most of the current underwater robots are artificial remote-controlled, which lack intelligent detection and autonomous grasping system. In this paper, a grasping robot equipped with an AI computing platform is developed to enable the autonomous grasping of underwater targets by using stereo vision technology. For the problem of difficult detection due to the small size and occlusion of underwater targets, this paper proposes Cascade DetNet, which can improve recognition accuracy. The experimental results show that our proposed method achieves the best performance on URPC dataset compared with several mainstream methods. In addition, we also carry out the autonomous grasping of seafood in a real marine environment to verify the autonomous grasping performance of underwater vehicles. [ABSTRACT FROM AUTHOR]

[Objectives] In this study, a method of detecting the external quality on freeze-dried strawberry based on binocular stereo vision was developed, in order to realize the nondestructive testing and grading of freeze-dried strawberry. [Methods]The binocular stereo vision technology was used to obtain the parallax data of each pixel in the freeze-dried strawberry image through the pixel matching algorithm, its three-dimensional model was established, and the volume of the model, color, shape and texture quality indexes of the image were calculated as the characteristic parameters. Then the successive projections algorithm (SPA)was used to screen the optimal characteristic parameters, and the external quality grade recognition model of freeze-dried strawberry was established with the unfiltered and filtered characteristic parameters respectively. In addition, the discrimination and grading effects of each model were compared. Finally, the optimal recognition model was selected to develop the external quality grading software of freeze-dried strawberry and verified. [Results]The total classification accuracy of the support vector machines (SVM) and the partial least squares discriminant analysis (PLS-DA) prediction models was 91.7% and 86.7%, respectively. The total classification accuracy of SVM model and PLS-DA model based on the feature parameters (volume and red component meanfiltered by SPA algorithm was 95.0% and 81.7% respectively. Finally, the external quality recognition software of freeze-dried strawberry was developed by using the SVM model constructed by the characteristic parameters screened by SPA algorithm, with the total accuracy of 90.8%. [Conclusions]The SVM hierarchical prediction model based on the feature parameters filtered by SPA had the better prediction effect, and the total accuracy of modeling set and prediction set were 97.8% and 95.0% respectively. In addition, the software developed based on the aforementioned model could realize the detection of the external quality of freeze-dried strawberry. [ABSTRACT FROM AUTHOR]

The development of superconducting magnetic bearings as the twisting element in ring spinning renews the study of the dynamic model of the ring spinning process. In this paper, the variational equation describing the steady motion of the yarn is developed using Hamilton's principle, the balloon shape is represented by the Bezier polynomial, the undetermined coefficient is solved by Galerkin method, and the semi-analytical solution of the steady-state balloon shape is obtained. Digital images of the balloon yarn are collected by binocular cameras and a stroboscope, and the balloon curve is reconstructed by stereo vision technology. The calculated balloon curve is in good agreement with the measured balloon curve. The balloon shapes and the radial forces acting on the permanent magnetic ring at different radii of the wind-point on the bobbin are investigated by the semi-analytical model when the spindle is spinning at different speeds. The solution procedures developed can be used to explore the vibration and stability of the system, consisting of the magnetic stiffness, permanent magnetic ring, and balloon yarn in the superconducting magnetic bearing twisting mechanism. [ABSTRACT FROM AUTHOR]

In recent years, the range of applications that utilize multiband imaging has significantly expanded. However, it is difficult to utilize multichannel heterogeneous images to achieve a spectral complementarity advantage and obtain accurate depth prediction based on traditional systems. In this study, we investigate CFNet, an iterative prediction network, for disparity prediction with infrared and visible light images based on common features. CFNet consists of several components, including a common feature extraction subnetwork, context subnetwork, multimodal information acquisition subnetwork, and a cascaded convolutional gated recurrent subnetwork. It leverages the advantages of dual-band (infrared and visible light) imaging, considering semantic information, geometric structure, and local matching details within images to predict the disparity between heterogeneous image pairs accurately. CFNet demonstrates superior performance in recognized evaluation metrics and visual image observations when compared with other publicly available networks, offering an effective technical approach for practical heterogeneous image disparity prediction.

The detection and localization of pineapple fruit must be successfully conducted to realize intelligent picking. This paper proposed a method for detecting and localizing pineapples in natural environments based on binocular stereo vision and an improved YOLO (You Only Look Once) v3 model. In comparison with the original YOLOv3, the improved algorithm did the following two improvement: the DenseNet was added into the Darknet-53 backbone network to optimize the 13 × 13 and 26 × 26 feature layer, and the SPP-net was fused in the 52 × 52 dimension detection module to strengthen the information representation ability of feature map. A binocular camera acquired left and right images. The left image was then input to the improved YOLOv3 model to obtain the position information of pineapples in the image. Stereo matching and parallax calculation of the target pineapple region were completed via the stereo matching algorithm. Finally, the three-dimensional co-ordinates of pineapples were calculated based on the triangulation principle of binocular stereo vision. A series of experiments were run to compare the detection result of this method against YOLOv3, Faster-RCNN (Region Convolutional Neural Network) and Mobilenet-SSD (Single Shot MultiBox Detector) as per their respective F1 score (Balanced Score) and AP (Average Precision) values. On test set with slight occlusion, the F1 score and AP values of the improved YOLOv3 model were 93.18% and 97.55%, respectively. As the occlusion grew severe, the F1 score and AP values decreased to 89.15% and 91.47%, respectively. The improved YOLOv3 model developed in this study had the best detection effect among all models tested. The binocular stereo vision localization experiment showed an average absolute error of 24.414 mm and average relative error of 1.17% at a distance of 1.7–2.7 m. The proposed method may thus be suitable for picking robots detecting and localizing pineapple fruit in natural environments. [ABSTRACT FROM AUTHOR]

Background: Camera calibration is a key process in optical deformation measurement techniques, like close-range photogrammetry and digital image correlation (DIC). Traditionally, an expensive calibration board with precisely machined feature points is employed in camera calibration. Low-precision calibration boards will lead to unstable camera parameters and worse length measurement accuracy (the relative measurement accuracy high than 0.1%). Objective: To introduce camera calibration methods capable of achieving high measurement accuracy even with low-precision calibration boards, and give practices guide on selecting suitable calibration methods and judging calibration results. Methods: Three calibration methods are introduced and discussed. In the iterative board-points method, all the feature points on the calibration board are flexible. The coordinates of all the feature points and camera parameters are iteratively updated and solved simultaneously. In the fixing 3-points method, three feature points on the calibration boards are selected and the distances between them are measured precisely; the locations of these points are fixed in the optimization process. Both the simulation experiments and the real experiments were carried out to study the performance of the three calibration methods. Results: Experimental results showed that both the iterative board-points method and the fixing 3-points method can effectively improve the accuracy of calculated camera parameters. Compared with the iterative board-points method, the fixing 3-points method restores the calibration board more precisely, as well as showing improved reproducibility and higher measurement accuracy. The relative accuracy of 3D distance measurement is also increased from 0.1% to 0.01 %, even using a low-precision paper calibration board. Conclusions: High measurement accuracy can be obtained using the fixing 3-points calibration method, even with a low-precision calibration board. This will decrease the cost and expand the application of optical measurement, especially in large structure deformation measurement. [ABSTRACT FROM AUTHOR]

Featured Application: The guidance of uncrewed aerial/ground vehicles, high-end security, surveillance, and various 3D manipulation, inspection, and measurements. Applying edge preservation filters for cost aggregation has been a leading technique in generating dense disparity maps. However, traditional approaches usually require intensive calculations, and their design parameters must be tuned for different scenarios to obtain the best performance. This paper shows that a simple texture-independent aggregation approach can achieve similar high performance. The proposed algorithm is equivalent to a sequence of matrix multiplications involving two weighting matrices and a primary matching cost. Notably, the weighting matrices are constant for image pairs with the same resolution. For higher matching accuracy, we integrate the algorithm with a multi-scale scheme to fully exploit the spatial distribution of textures in the image pairs. The resultant hybrid approach is efficient and accurate enough to surpass most existing approaches in stereo matching. The performance of the proposed approach is verified by extensive simulation results using the Middlebury (3rd Edition) benchmark stereo database. [ABSTRACT FROM AUTHOR]

Intelligent mining needs to achieve real-time acquisition of surrounding rock deformation data of roadways and analysis and provide technical support for intelligent mining construction. To solve problems such as significant error, information lag, and low acquisition frequency of surrounding rock monitoring technology at the current stage, a perception method, RSBV of roadway deformation situation, based on binocular vision is proposed, which realizes the dynamic, accurate real-time acquisition of host rocks' relative deformation in a dusky environment. The low illumination image enhancement method is used to preprocess original images, which reduces the impact of low illumination and high dust; the K-medoids algorithm segments the target image, and the SIFT algorithm extracts feature points from the ROI (region of interest). The influence of eliminating background images on the feature point extraction is revealed, and the efficiency of feature extraction is improved; the method of SIFT feature-matching with epipolar constraints is studied, which improves the accuracy and speed of feature points. The roadway deformation characteristics are analyzed, and the reflective target is used as the monitoring point. A roadway deformation acquisition and analysis platform based on binocular vision is built in a dim environment. Zhang's method is selected to calibrate the camera parameters, and stereo rectification is carried out for the target motion image. The adaptability of the RSBV method to different surrounding rock deformation scales is studied and compared with the measurement results of the SGBM algorithm. The results show that the error of the RSBV method is controlled within 1.6%, which is 2.88% lower than the average error of the SGBM algorithm. The average time for processing a group of binocular images is 1.87 s, which is only 20% of the SGBM algorithm. The research result provides a reliable theoretical basis for the real-time and accurate evaluation of the surrounding rock deformation mechanism. [ABSTRACT FROM AUTHOR]

At present, there are still problems of inaccurate labeling, inaccurate target matching, and excessively complex image processing of machine vision in spatial structure displacement measurement. Therefore, machine vision measurement technology is combined with deep learning algorithm to construct a precise displacement measurement model for largespan spatial structure of buildings based on machine vision technology and deep learning. First. the binocular stereo vision technology based on machine vision is expounded to analyze the deficiencies in measuring displacement of large-scale spatial structures. Second. related concepts of Convolutional Neural Networks (CNNs) are introduced, and You Only Look Once v2 (YOLOv2) is selected as the recognition algorithm of sphere joints iii the long-span spatial structure of buildings. Finally. a displacement measurement system for long-span spatial structures is constructed based on CNN models and YOLOv] algorithm. Images of various spatial structures collected through the Internet and shot by smartphones constitute the data set to test the performance of YOLOv2 recognition algorithm and three kinds of CNN models, namely the AlexNet model, Darkneti 9 model, and Resnet] 52 model. The experimental results demonstrate that the Recall of Darknet19 model is up to 93.9%, and the Intersection Over Union (IOU) is 82.02%, which are at least 3.6% and 4.4% higher than those of the other two niodels, respectively. Besides, the Darknet]9 model performs better in the task of identifying and locating sphere joints in spatial structures. In addition. when the Darknet 19 model is trained about 20,000 times, the Recall of the YOLOv2 recognition algorithm reaches 94.69%, and the IOU attains 84.98 %. Moreover, the average recognition accuracy of YOLOv2 algorithm is 2.3% higher than other recognition algorithms, which has a high recognition level. Furthermore. the relative error of displacement measurement in the single direction of each stage of the measurement modeldesigned here is controlled within 12%, which can meet the requirements of displacement measurement of spatial structure. The purpose of this work is to provide essential technical support for the upgrading of displacement measurement technology of large-span spatial structure of buildings and the improvement of measurement accuracy. [ABSTRACT FROM AUTHOR]

[Display omitted] • An innovative deep learning-based DAEV-Stereo model is proposed for the stereo matching of coal slurry froth images. • Modelling of foam stereo images to create virtual foam image datasets. • This study extracted the three-dimensional attributes of froth images. This study focuses on the parameters necessary for the successful deployment of intelligent control systems for flotation processes, focusing on froth surface characteristics. Depth information is not available for conventional 2D features extracted from single froth images, thereby constraining their capacity to fully capture froth surface characteristics. To overcome this limitation, this study proposes a methodology for extracting 3D froth features by employing a deep learning-based binocular stereo vision approach. Initially, the froth image is preprocessed, including correction and segmentation. Subsequently, a froth stereo matching model called dual-attention encoding volume stereo (DAEV-Stereo) is developed. This model is trained utilizing simulated and real-world datasets to determine the froth parallax. The distance between the froth surface and camera is computed through the binocular vision approach based on camera intrinsics and extrinsics. Subsequently, the Froth is reconstructed in three dimensions, and the surface area and volume are calculated utilizing the triangular prism differential traversal technique. A mathematical model is formulated to integrate depth information and froth layer thickness to compute colour characteristics. The experimental results indicate that the DAEV-Stereo model obtains an endpoint error of 0.5 for froth stereo matching and a processing speed of 0.37 s, thereby satisfying the operational criteria for industrial use. After conducting pilot plant flotation tests, the median absolute deviations for the froth surface area, volume, and colour features were determined as 18.42, 17.62, and 9.67, respectively. Moreover, compared with traditional features, stereo features demonstrate reduced fluctuations, improved stability, and better correlations with operational conditions. The binocular stereo vision approach and DAEV-Stereo model for extracting 3D froth features are valuable for accurately characterizing froth surface data. [ABSTRACT FROM AUTHOR]

• A stereo matching algorithm for 3D reconstruction of mineral images is proposed. • The proposed BT-Census algorithm framework is developed by Semi-Global-based method. • The matching cost is calculated by fusing Gaussian weighted Census cost and BT cost. • An adaptive window cost aggregation and a four-way scanline optimization are used. • Experiments indicate that the matching rate and accuracy are improved significantly. Binocular stereo matching is crucial for identifying and locating minerals on the grizzly, allowing the robotic system to carry out crushing autonomously. The traditional stereo matching algorithm yields a low matching rate due to the relatively single color and weak texture of the mineral image caused by the uneven illumination in the field. An improved Birchfield-Tomasi (BT)-Census algorithm is proposed to enhance the capability of discriminating the mineral region and increase the successful matching rate. Firstly, the Gaussian-weighted average grey value of the circular window is used as the central value of the Census transform, and the initial surrogate value is obtained by weighting and fusing the Census cost and the BT cost. Subsequently, the cost aggregation method by adaptive windows is used, and then scanline optimization is applied to select the optimal matching cost. The performance evaluation results using the Middlebury dataset show that the proposed algorithm achieves a 93.33% average successful matching rate, outperforming Absolute Difference of Intensity (AD)-Census, Semi-Global Matching (SGM), and PatchMatch algorithms by 6.5%, 8.04%, and 4.62% respectively. Moreover, In the three-dimensional (3D) reconstruction experiments of minerals on grizzly, the point cloud reconstructed by the proposed method shows significant improvement in terms of accuracy. Notably, in comparison to the SGM algorithm, there is an 83.4% reduction in Mean-Square Error (MSE), a 35.4% reduction in Root Mean-Square Error (RMSE), and a 35.8% reduction in Mean Absolute Error (MAE). Against the AD-Census algorithm, reductions of 47.8% in MSE, 21.6% in RMSE, and 21.4% in MAE are observed. Similarly, in comparison to the PatchMatch algorithm, there are reductions of 11.9% in MSE, 5.8% in RMSE, and 6.1% in MAE. In a word, the proposed improved BT-Census stereo matching algorithm effectively enhances the detailed features of the minerals and improve the successful matching rate and accuracy. [ABSTRACT FROM AUTHOR]

• Identifying the correspondence of multi-line structured light in binocular images. • Refining the matching light stripes via coarse-to-fine spatial geometric constraints. • The proposed method needs no coding scheme, and has high flexibility and speed. • The recognition accuracy of the proposed method in different scenarios reaches 99.4%. In the 3D reconstruction of binocular stereo vision with multi-line structured light, the structured light projections in the 2D image may become distorted, cracked, or disappear due to the modulation of the surface of the object being measured. Correctly distinguishing the light plane to which the laser stripe belongs from the image is a fundamental and challenging problem in structured light binocular vision. This has a direct impact on the effectiveness of reconstructing 3D information. In this paper, we solve this problem via coarse to refined spatial geometric constraints. Firstly, a pre-location method for laser stripe points based on the perspective projection relationship is proposed. The method coarsely identifies the relationship between the corresponding laser stripes in the binocular image and the light plane to which they belong. Then, the minimum distance constraint from the point to the light plane is employed to refine the correlation between the corresponding laser stripes in the binocular image. Finally, the proposed method achieves an average accuracy of 99.4 % when experimentally analyzing different measured objects. The method has been implemented on a GPU using CUDA, with an average processing time of 35 ms per frame. [ABSTRACT FROM AUTHOR]

(1) Background: In real road scenarios, various complex environmental conditions may occur, including bright lights, nighttime, rain, and snow. In such a complex environment for detecting pole-shaped obstacles, it is easy to lose the feature information. A high rate of leakage detection, false positives, and measurement errors are generated as a result. (2) Methods: The first part of this paper utilizes the improved YOLOv5 algorithm to detect and classify pole-shaped obstacles. Then, the identified target frame information is combined with binocular stereo matching to obtain more accurate distance information. (3) Results: The experimental results demonstrate that this method achieves a mean average precision (mAP) of 97.4% for detecting pole-shaped obstacles, which is 3.1% higher than the original model. The image inference time is only 1.6 ms, which is 1.8 ms faster than the original algorithm. Additionally, the model size is only 19.0 MB. Furthermore, the range error of this system is less than 7% within the range of 3–15 m. (4) Conclusions: Therefore, the algorithm not only achieves real-time and accurate identification and classification but also ensures precise measurement within a specific range. Meanwhile, the model is lightweight and better suited for deploying sensing systems.

In this paper, using automated optical inspection equipment and a thermal imager, the position and the temperature of the heat source or measured object can effectively be grasped. The high-resolution depth camera is with the stereo vision distance measurement and the low-resolution thermal imager is with the long-wave infrared measurement. Based on Planck's black body radiation law and Stefan–Boltzmann law, the binocular stereo calibration of the two cameras was calculated. In order to improve the measured temperature error at different distances, equipped with Intel Real Sense Depth Camera D435, a compensator is proposed to ensure that the measured temperature of the heat source is correct and accurate. From the results, it can be clearly seen that the actual measured temperature at each distance is proportional to the temperature of the thermal image vanadium oxide, while the actual measured temperature is inversely proportional to the distance of the test object. By the proposed compensation function, the compensation temperature at varying vanadium oxide temperatures can be obtained. The errors between the average temperature at each distance and the constant temperature of the test object at 39 °C are all less than 0.1%. [ABSTRACT FROM AUTHOR]

Obstacle detection and recognition are core problems in the field of autonomous driving. For urban road surfaces, this paper proposes a complete obstacle recognition method. Firstly, this paper differs from semantic segmentation and target detection frameworks but represents the region of interest based on stixel and improves the traditional scheme of generating stixel. Then this paper classifies the different obstacles in the region of interest based on the improved deep learning network DenseNet. Finally, recognition of individual obstacles is performed. The experimental results show that the method in this paper can effectively represent the region of interest consisting of the first layer of obstacles and achieve obstacle recognition more accurately. [ABSTRACT FROM AUTHOR]

At present, the weighing process of concrete aggregate is cumbersome, and the weighing result has a large error. This paper proposes a method for measuring the weight of irregular bone stockpiles based on binocular vision. Firstly, based on the principle of binocular vision, the improved SMG stereo matching algorithm and OTSU threshold segmentation algorithm are used to calculate the 3D point cloud coordinates corresponding to the disparity map. Then, the 3D reconstruction of the aggregate pile and the calculation method of volume and mass are proposed. The results show that the method can accurately measure the aggregate pile weight, and the measurement error is less than 6%. The method can effectively reduce the operating cost of the concrete batching station and improve the measurement efficiency. [ABSTRACT FROM AUTHOR]

Liquified natural gas (LNG) manipulator arms have been widely used in natural gas transportation. However, the automatic docking technology of LNG manipulator arms has not yet been realized. The first step of automatic docking is to identify and locate the target and estimate its pose. This work proposes a petroleum pipeline interface recognition and pose judgment method based on binocular stereo vision technology for the automatic docking of LNG manipulator arms. The proposed method has three main steps, including target detection, 3D information acquisition, and plane fitting. First, the target petroleum pipeline interface is segmented by using a color mask. Then, color space and Hu moment are used to obtain the pixel coordinates of the contour and center of the target petroleum pipeline interface. The semi-global block matching (SGBM) algorithm is used for stereo matching to obtain the depth information of an image. Finally, a plane fitting and center point estimation method based on a random sample consensus (RANSAC) algorithm is proposed. This work performs a measurement accuracy verification experiment to verify the accuracy of the proposed method. The experimental results show that the distance measurement error is not more than 1% and the angle measurement error is less than one degree. The measurement accuracy of the method meets the requirements of subsequent automatic docking, which proves the feasibility of the proposed method and provides data support for the subsequent automatic docking of manipulator arms. [ABSTRACT FROM AUTHOR]

Based on binocular stereo vision measurement and wireless sensing theory, this paper studies the measurement and analysis methods of athletes' movement displacement parameters. The main purpose is to explore the relationship between sports biomechanical parameters. The analysis reveals the laws of athletes 'movements, and provides theoretical guidance on athletes' attitude recognition and physical training. Based on the principle of binocular stereo vision, a non-contact displacement measurement system was developed. The displacement measurement system uses template matching to extract the image coordinates of the measurement points and restore the spatial coordinates of the measurement points through European 3D reconstruction, which improves the practicality of the measurement system. In the cantilever beam vibration test, the traditional displacement sensor and binocular image displacement measurement system are used to simultaneously measure the cantilever beam displacement. The experimental results show that the integrated displacement measurement system in this paper has higher accuracy and better stability.

To achieve the rapid recognition and accurate picking of Camellia oleifera fruits, a binocular vision system composed of two industrial cameras was used to collect images of Camellia oleifera fruits in natural environments. The YOLOv7 convolutional neural network model was used for iterative training, and the optimal weight model was selected to recognize the images and obtain the anchor frame region of the Camellia oleifera fruits. The local binary pattern (LBP) maps of the anchor frame region were extracted and matched by using the normalized correlation coefficient template matching algorithm to obtain the positions of the center point in the left and right images. The recognition experimental results showed that the accuracy rate, recall rate, mAP and F1 of the model were 97.3%, 97.6%, 97.7% and 97.4%. The recognition rate of the Camellia oleifera fruit with slight shading was 93.13%, and the recognition rate with severe shading was 75.21%. The recognition rate of the Camellia oleifera fruit was 90.64% under sunlight condition, and the recognition rate was 91.34% under shading condition. The orchard experiment results showed that, in the depth range of 400–600 mm, the maximum error value of the binocular stereo vision system in the depth direction was 4.279 mm, and the standard deviation was 1.142 mm. The detection and three-dimensional positioning accuracy of the binocular stereo vision system for Camellia oleifera fruits could basically meet the working requirements of the Camellia oleifera fruit-picking robot.

To improve the navigation safety of inland river ships and enrich the methods of environmental perception, this paper studies the recognition and depth estimation of inland river ships based on binocular stereo vision (BSV). In the stage of ship recognition, considering the computational pressure brought by the huge network parameters of the classic YOLOv4 model, the MobileNetV1 network was proposed as the feature extraction module of the YOLOv4 model. The results indicate that the mAP value of the MobileNetV1-YOLOv4 model reaches 89.25%, the weight size of the backbone network was only 47.6 M, which greatly reduced the amount of computation while ensuring the recognition accuracy. In the stage of depth estimation, this paper proposes a feature point detection and matching algorithm based on the ORB algorithm at sub-pixel level, that is, firstly, the FSRCNN algorithm was used to perform super-resolution reconstruction of the original image, to further increase the density of image feature points and detection accuracy, which was more conducive to the calculation of the image parallax value. The ships' depth estimation results indicate that when the distance to the target is about 300 m, the depth estimation error is less than 3%, which meets the depth estimation needs of inland ships. The ship target recognition and depth estimation technology based on BSV proposed in this paper makes up for the shortcomings of the existing environmental perception methods, improves the navigation safety of ships to a certain extent, and greatly promotes the development of intelligent ships in the future. [ABSTRACT FROM AUTHOR]