231 results about "Vision Disparity" patented technology

The invention particularly relates to a binocular vision-based three-dimensional reconstruction method, which comprises the following steps of: calibrating a binocular camera based on an improved Zhang Zhengyou calibration method to obtain internal and external parameters and a distortion coefficient of the camera; performing stereo correction on the image by using the internal and external parameters of the camera and the distortion coefficient obtained by calibration, so that the binocular image meets an epipolar constraint condition; a multi-strategy optimized semi-global stereo matching algorithm is adopted to process the image after stereo correction, and a disparity map is generated; based on the generated disparity map, generating a three-dimensional point cloud through a triangulation principle; carrying out anti-interference processing and registration optimization on the three-dimensional point cloud; and performing global splicing on the three-dimensional point clouds subjected to anti-interference processing and registration optimization based on a sequential registration error sharing strategy to complete three-dimensional reconstruction. According to the method, the key problems of large calibration error, weak texture matching failure, point cloud noise sensitivity and registration accumulative error in a traditional method are solved, and the reconstruction precision and stability are remarkably improved.

The invention relates to the technical field of image data processing and intelligent control, and discloses a patch board spot welding track intelligent control method and system based on visual perception. According to the method, synchronous image streams are collected through a binocular vision sensor, and epipolar correction image pairs are generated through timestamp synchronization, ROI extraction and epipolar correction processing; calculating a disparity map by adopting a stereo matching algorithm, and constructing a compensated three-dimensional point cloud model based on deformation vector field fusion multi-frame point cloud data of a radial basis kernel function; welding spot position error vectors are generated by extracting welding spot feature points and performing spatial filtering optimization; and carrying out inverse kinematics calculation on the mechanical arm by adopting a damping least square method, carrying out safety constraint optimization in combination with prospective collision risk assessment and real-time pose data, and generating a trajectory compensation instruction. According to the method, the problems of dynamic deformation compensation and motion safety in patch plate welding are solved, the submillimeter welding spot positioning precision is achieved, and the welding quality stability and the system robustness are improved.

The invention provides a robust stereo matching method fusing monocular semantic prior and multi-dimensional expert aggregation, relates to the technical field of computer vision and stereo matching, establishes a robust stereo matching framework fusing monocular semantic prior and multi-dimensional expert aggregation, and comprises a monocular branch and a binocular branch, inputting a stereo image into a robust stereo matching framework based on fusion of monocular semantic prior and multi-dimensional expert aggregation for matching calculation, wherein the matching calculation comprises the following steps: extracting semantic prior features of the input image by using a monocular branch; the method comprises the following steps: extracting a geometric enhancement feature map of a stereo image by using binocular branches, generating a matching cost body based on the geometric enhancement feature map, and carrying out refined iterative updating on the matching cost body to obtain a matched disparity map. According to the method, monocular and binocular depth estimation is cooperatively supported in a unified network architecture, and a parallel monocular depth prior path is utilized to actively guide and strengthen a core binocular matching process.

The invention relates to the technical field of intelligent maintenance of tunnel electromechanical facilities, in particular to a tunnel lamp recognition and spatial positioning method based on binocular vision, which comprises the following steps: synchronously acquiring tunnel environment images through left and right cameras subjected to polar horizontal correction, inputting the images into a lightweight tunnel lamp detection model, and detecting an area image containing a lamp bounding box; extracting feature points in the bounding box by adopting a multi-scale adaptive FAST feature extraction algorithm, describing the feature points by rotating an invariant binary descriptor, and generating a disparity map in combination with an adaptive Hamming distance matching algorithm; calculating three-dimensional space coordinates of the lamp according to the disparity map by using a binocular camera triangulation principle and combining camera parameters; when the ambient illumination is lower than a threshold value, a multispectral image fusion module is started to enhance the feature matching robustness; and when continuous matching fails, visible light communication assisted positioning is started. According to the invention, the problem of low positioning precision of the lamp in a complex tunnel environment is solved, and the reliability and efficiency of identification and positioning are improved.

The invention discloses a photoelectric fusion beam control method and device based on a reconfigurable intelligent reflecting surface. The method comprises the following steps: acquiring internal and external parameter matrixes of a binocular camera and a binocular-RIS coordinate transformation matrix; capturing a binocular image stream of a detection target by using a binocular camera and executing stereo matching to obtain a disparity map and depth data of a binocular image; detecting a target image in the binocular image and outputting bounding box information of a center point of the target image; calculating target image coordinates through the depth data and the bounding box information, and converting the target image coordinates into three-dimensional space coordinates under an RIS coordinate system so as to determine the spatial position of the target; generating an RIS phase codebook according to the target spatial position, wherein the RIS phase codebook comprises a near-field phase compensation item; and loading the phase codebook to the RIS unit to form a target beam so as to perform directional tracking on a target. According to the technical scheme of the invention, high-precision beam control and stable positioning and tracking of the target can be realized in a complex multi-target environment.

The invention provides a short-baseline binocular three-dimensional human body posture reconstruction method and system, and the method comprises the steps: collecting original left and right image pairs, and carrying out the detection of a human body region; carrying out human body image pair matching and scale transformation on the detected human body region; parallax features are obtained based on the processed left and right human body area image pairs, feature fusion is carried out on the parallax features and human body features extracted by the three-dimensional human body posture estimation sub-network, a parallax image initial result is obtained based on feature fusion information, a loss function is established with the human body features to carry out parallax optimization, and an optimized parallax image and a scene depth image are obtained; human body features are obtained based on left and right human body area image pairs, feature fusion is carried out on the human body features and parallax features, two-dimensional human body joint points are obtained through two-dimensional posture estimation according to feature fusion results, and a three-dimensional human body posture initial result is obtained based on binocular camera parameters and a triangulation calculation method; and establishing a loss function with the scene depth map and human body prior supervision, and performing three-dimensional human body posture optimization to obtain an optimized three-dimensional human body posture.

The invention belongs to the technical field of data measurement, and discloses a high-speed railway ballastless track construction measurement method, which comprises the following steps of: carrying out internal reference and external reference calibration on a binocular camera of a track detection trolley; obtaining related data containing a binocular image, and carrying out space-time alignment processing on the related data so as to construct a prediction state vector; carrying out stereo matching on the binocular image to generate a prediction disparity map, and carrying out three-dimensional reconstruction on the prediction disparity map to obtain a track point cloud map; and extracting a track inner side point coordinate sequence and a track height program sequence from the track point cloud picture, and generating a corresponding track gauge table and a corresponding smoothness table. According to the invention, vision, three-dimensional reconstruction and inertial navigation technologies are integrated, and integrated detection of gauge, elevation and smoothness parameters is realized.

The invention discloses a pavement disease size accurate quantification method based on binocular vision, and the method specifically comprises the steps: collecting pavement image data through a vehicle-mounted binocular camera, achieving the real-time detection of a disease target through an improved RT-DETR model, and outputting the disease type and detection frame information; after a pavement disease target is detected, pixel-level segmentation is carried out on a disease area in the detection frame based on a semantic segmentation model, and a disease contour mask is extracted; an improved IGEV-Stereo stereo matching algorithm is used for calculating a disparity map, depth information is output based on camera calibration parameters and a disparity estimation result, and conversion from pixel coordinates to three-dimensional coordinates is achieved; and the binocular depth information and a disease detection segmentation result are combined to realize accurate size quantification of typical road surface diseases with different characters. Through the binocular stereoscopic vision technology, automatic detection and precise quantification of pavement diseases can be realized, the cost can be effectively reduced while the disease quantitative evaluation precision is improved, and decision support is provided for road management and maintenance.

The invention relates to the technical field of computer vision, and particularly provides a tunnel video stream three-dimensional modeling system based on binocular stereo matching and SLAM (Simultaneous Localization and Mapping), which comprises a multi-modal image acquisition unit, a binocular infrared camera and an RGB (Red, Green and Blue) camera are configured to synchronously acquire an infrared image pair, an RGB image pair and inertial measurement unit data of a tunnel environment, the multi-source data is uploaded to the cloud processing platform in real time through the wireless transmission module; based on a dynamic switching mechanism of environmental perception, performing adaptive fusion pose estimation of a feature point method and a direct method on input multi-modal image data; the stereo matching and dense reconstruction unit is used for calculating a disparity map by adopting an improved self-adaptive window stereo matching algorithm, generating a three-dimensional point cloud in combination with the pose information output by the pose estimation unit, and constructing a global consistency point cloud model of the tunnel scene through a time sequence point cloud registration and splicing algorithm; and a multi-algorithm target detection and semantic fusion unit. The method can meet the requirements of precision and robustness of tunnel reconstruction.

The invention relates to the technical field of computer vision, in particular to a three-dimensional reconstruction method and device based on optical polarization and stereoscopic vision principles. The three-dimensional reconstruction method sequentially comprises the steps of image preprocessing, polarization parameter extraction, binocular stereo matching, disparity map generation and optimization, normal vector correction, depth information fusion and three-dimensional point cloud reconstruction and modeling generation. A polarization normal vector field is constrained and corrected as a'skeleton ', and the azimuth angle pi ambiguity problem which puzzles polarization three-dimensional reconstruction for a long time is solved; high-frequency surface normal details contained in corrected polarization information are used as textures to enhance and fill up depth information loss of binocular vision in weak texture and repeated texture areas, the inherent limitation of a single sensing technology in a three-dimensional reconstruction task is overcome, and the three-dimensional reconstruction of the surface of an object, especially a diffuse reflection object, is realized. And high-precision and high-integrity three-dimensional shape recovery is realized.

The invention provides a three-dimensional calibration method using a speckle image as a calibration object and combining weight parameters. The method comprises the following steps: acquiring a standard grid image and the speckle image through a camera; determining angular point positions in the standard grid image through an angular point detection algorithm; calibrating the stereoscopic vision system by using the obtained angular points to obtain an initial calibration result; calculating 3D coordinates of the camera calibration board in the world by using the disparity map; determining a subset of the reference image corresponding to the checkerboard angular points in the target image through a digital image cross-correlation algorithm; calculating a weight parameter of each calibration point based on the re-projection error; solving internal and external parameters of the camera by using the weight parameters, radial alignment constraint and a least square method; optimizing the main point of the camera by using the weight parameters; and if the re-projection error or the number of iterations does not meet the requirement, repeatedly calculating the weight parameter of the calibration point and the subsequent steps, otherwise, obtaining a camera calibration result, and obtaining the distortion curved surface of the camera lens.

The invention provides a super-resolution binocular image generation method and system based on geometric structure consistency, and the method comprises the steps: extracting the deep features of a low-resolution binocular image through employing a convolutional neural network or a Transform model, achieving the information interaction of a left image and a right image in combination with a cross attention module, and constructing a pixel incidence matrix of the left image and the right image; acquiring a pixel corresponding relation of the left image and the right image by using the pixel incidence matrix, and constructing a continuous parallax field; performing spatial warping on the deep features based on a continuous parallax field to obtain warping features, and aligning the deep features of the left and right images; and merging the deep features and the warping features after spatial alignment, inputting the merged features into a feature up-sampling module based on implicit two-dimensional expression, and outputting a high-resolution binocular image of the same scene. The invention provides a binocular image super-resolution technology comprising binocular image feature extraction, continuous parallax field construction based on implicit two-dimensional expression, left and right image feature space alignment and feature upsampling based on implicit two-dimensional expression.

The invention discloses an underwater fish body length measurement method, system and device based on binocular vision and a medium, and relates to the technical field of underwater measurement. The method comprises the following steps: performing fish body tracking on a collected fish school video, and determining an ID of each fish and corresponding position information; judging the straightening state based on the tracking frame corresponding to the ID of each fish, and when the judgment result is that the fish body is straightened, selecting a frame with the longest center line as a fish body straightening reference frame; carrying out ROI cutting and masking on the basis of the fish body straightening reference frame to obtain an ROI area map; inputting the left view and the right view in the ROI into a GMFlow model for parallax matching to obtain a parallax image which only contains the characteristics of the whole body of the target fish and the periphery of which forms a black frame; and calculating the length of the fish body according to the coordinates of the central points of the four corners of the fish body in the disparity map and calibration parameters of the binocular camera. According to the invention, efficient and accurate fish body measurement can be realized through step-by-step processing of the fish school image sequence.

The invention provides a road three-dimensional lane line detection method and system based on binocular vision, and relates to the technical field of automatic driving, and the method comprises the steps: carrying out the online calibration of internal and external parameters of a camera and the time sequence alignment of the binocular image and IMU attitude data through collecting the forward binocular image and IMU attitude data of a vehicle; processing the binocular image after time sequence alignment, and determining a dense disparity map; generating a road point cloud based on the dense disparity map and camera parameters, and constructing an adaptive terrain model; constructing a depth enhanced BEV feature map according to the depth feature of the road point cloud and the texture feature of the binocular image, predicting candidate parameters of a three-dimensional lane line, and screening through various constraints; and then executing extended Kalman filtering and trajectory optimization to obtain three-dimensional lane line parameters. According to the invention, rapid and accurate detection of the three-dimensional lane line can be realized, the accuracy of three-dimensional lane line detection under a complex terrain is improved, and the deployment cost is reduced.

The application provides a three-dimensional reconstruction method based on binocular stereo matching, and relates to the technical field of binocular stereo vision. The method combines binocular stereo matching algorithm, triangulation algorithm and surface texture mapping, and can realize surface reconstruction of scene objects in different environments. After stereo calibration is performed on a binocular camera, the internal and external parameters of the binocular camera are obtained, stereo correction is performed, then an image disparity map of the corresponding scene object is generated according to an optimized semi-global stereo matching algorithm, after the disparity information is obtained, the point cloud information converted by the object disparity is triangulated on the surface according to the triangulation algorithm, and finally three-dimensional surface reconstruction of the object is realized through the texture mapping technology. The application can reduce the influence of noise in the disparity map generation process, i.e. the semi-global stereo matching algorithm, improve the accuracy of the disparity information, and complete the three-dimensional surface reconstruction task of the object with little sacrifice of time performance.

The invention discloses a distance measuring method and device based on binocular vision and laser, and intelligent wearable equipment. The method comprises the following steps: collecting a first image and a second image of a target object at different visual angles through a binocular vision device of the intelligent wearable equipment; extracting, matching and screening feature points of a target object based on the first image and the second image to obtain key point pairs, and calculating parallax according to the key point pairs; calculating a depth value of the intelligent wearable device and the target object according to the parallax; laser ranging values of the intelligent wearable device and the target object are collected through a laser device of the intelligent wearable device; fusing the depth value and the laser ranging value by adopting a Kalman filtering algorithm, and outputting a fused ranging result; the method solves the problems that a distance measurement scheme in the prior art is complex in algorithm and high in cost.

The embodiment of the invention discloses an aircraft height measurement method based on binocular vision, and the method comprises the steps: obtaining internal parameters and distortion parameters of a first camera and a second camera, and obtaining a first image and a second image collected by the first camera and the second camera; calibrating the first image and the second image according to the internal reference and the distortion parameter of the first camera and the second camera to obtain a first calibration image and a second calibration image; binocular matching is carried out on the first calibration image and the second calibration image, and a full-image disparity map is obtained based on the first calibration image; performing airport runway semantic segmentation based on the first calibration image to obtain a semantic segmentation map of the airport runway; performing 3D coordinate reduction operation on the 2D pixels according to the full-image disparity map and the semantic segmentation map to obtain a 3D point cloud of the airport runway; fitting a runway plane equation according to the 3D point cloud, converting the runway plane equation to an airframe coordinate system according to the relative position of the first camera and the airframe, and obtaining an aircraft height measurement result according to the runway plane equation under the airframe coordinate system.

The invention relates to the field of binocular stereo vision, and discloses a binocular stereo matching method and system based on multi-scale iterative optimization and related equipment. The method comprises the steps of performing semantic structure feature extraction on a corrected left view and a corrected right view through a feature extraction network; determining a cost space pyramid and a probability matrix according to the left view feature map and the right view feature map output by the feature extraction network; extracting a multi-scale context feature of the left view through a context sensing network, and initializing the multi-scale context feature to obtain a hidden state and input of a recursive network; during recursive network iteration, according to the probability matrix, a multi-range search strategy is adopted to index local cost from the cost space pyramid; and inputting the local cost into the recursive network, combining the hidden state obtained by the context features and the input, and carrying out iterative updating on the parallax value to obtain a parallax map, thereby improving the accuracy of parallax calculation.

The invention discloses an indexable blade cutting edge passivation radius measuring method based on binocular vision, and belongs to the technical field of binocular vision measurement. The method comprises the following steps: calibrating left and right cameras by using a binocular vision system to obtain internal and external parameters and distortion coefficients of the cameras; secondly, acquiring an indexable blade cutting edge passivation image for image correction, so that the corrected images are in the same plane and are parallel to each other; secondly, improving the blade image quality through a wavelet packet transformation image preprocessing method, and reducing the noise influencing the image quality; thirdly, obtaining an image disparity map through a region-based optimization SGBM stereo matching algorithm, and changing the disparity map into a depth map according to a triangulation principle; and finally, measuring the passivation radius of the cutting edge of the blade on the basis of the obtained depth map, and comparing the passivation radius of the cutting edge of the blade with the passivation radius of the cutting edge of the blade measured by an Alcona three-dimensional detection instrument so as to judge the precision of the research.

The invention discloses a two-dimensional detection and SGBM three-dimensional ranging method based on improved YOLOv13, belongs to the field of computer vision, and particularly relates to the two-dimensional detection and SGBM three-dimensional ranging method based on the improved YOLOv13. The objective of the invention is to solve the problem of low cross-view-angle distance measurement precision in a three-dimensional space perception level of the existing method. The method comprises the following steps of: obtaining a trained YOLOv13-SHSA network model; the binocular camera obtains a left view and a right view; inputting the left view into the trained network model, and outputting three detection results by three detection heads of the trained network model; processing the three detection results to obtain detection frames and confidence coefficients of a plurality of targets in the left view; obtaining a disparity map of the left and right views; calculating a depth value corresponding to each pixel point in the disparity map according to the disparity map and the binocular camera parameters; and calculating the distance of the target object based on the depth map, the detection frames of the multiple targets in the left view and the confidence coefficient.

The invention discloses a binocular fisheye stereo matching method constructed by fusing monocular depth prior and multi-scale spherical projection cost. The binocular fisheye stereo matching method comprises the following specific steps: S1, image input and feature extraction; s2, monocular depth priori estimation; s3, constructing a cost matrix of multi-scale spherical projection scanning; s4, the deep attention network is matched; s5, performing initial parallax estimation; s6, obtaining a final fine disparity map based on a ConvGRU disparity optimization network, and the method can accurately and efficiently carry out stereo matching without image correction, and improves the matching precision and enhances the robustness of an algorithm to a distorted image by introducing monocular depth estimation prior and a multi-scale spherical projection scanning strategy; the method also avoids the quality loss caused by image correction, retains the original view field information of the fisheye image, gives consideration to the precision and efficiency through a multi-scale spherical scanning strategy, adapts to fisheye distortion characteristics, introduces an attention mechanism and a GRU structure, and improves the expression capability of a parallax estimation network.

The invention belongs to the technical field of underwater target space positioning, and discloses an underwater benthos rapid positioning method and system based on binocular vision, and the system comprises an underwater environment data binocular camera collection device which carries out the image enhancement processing of collected image data; inputting the enhanced image into a lightweight Slim-RT-DETR target detection network, and carrying out the detection and recognition of a fishing object; cutting target areas of left and right views of the binocular camera based on a target detection and recognition result, and calculating a parallax value by adopting an anti-noise optimization method of random sampling and neighborhood interpolation; and calculating three-dimensional space coordinates of the fishing object according to the binocular parallax and the re-projection matrix, and positioning the underwater benthos. According to the invention, a rapid image enhancement algorithm and a target detection model suitable for underwater fishing are designed, a normal form method suitable for underwater biological positioning is provided, and high-precision and high-real-time underwater fishing object positioning can be realized.

The invention discloses a three-dimensional reconstruction and pose estimation system and method based on binocular structured light, and the method comprises the steps: obtaining binocular structured light image pairs and attitude angles of a detected part at different preset attitude angles, and synchronously binding the attitude angles with the binocular structured light images; reconstructing dense three-dimensional point clouds of the surface of the measured part in each attitude by utilizing parallax information between binocular images and combining structured light coding and triangulation principles; performing initial pose estimation by using an efficient multi-view algorithm, and performing registration on the dense three-dimensional point cloud under each pose by using a multi-scale geometric constraint ICP algorithm based on an initial pose estimation result and the pose angle to obtain a registered complete point cloud; and geometric completion is carried out on sparse point clouds in the obtained registered complete point clouds through a PCN model, and a global complete three-dimensional point cloud model of the measured part is generated. The method has good precision and robustness in reconstruction of the complex special-shaped part, and the system is light, rapid in deployment and stable in measurement.

The invention belongs to the field of computer graphics, and discloses a self-adaptive LOD subdivision method and system oriented to virtual reality and based on a perceptual model. The method comprises the following steps: firstly, researching factors influencing visual perception, establishing a perception model BD-castleCSF fused with stereoscopic vision by combining binocular parallax, and dynamically dividing a rendering region; then, on the basis of a discrete LOD initial order processing grid model, introducing a mosaic technology to execute grid subdivision, calculating a downsampling factor according to a perception model, dynamically adjusting the subdivision degree, and proposing a perception adaptive subdivision LOD algorithm; experimental results show that compared with a traditional static rendering method, the LOD switching frequency of the method is reduced by 67%, the effective subdivision triggering rate is improved by 19.9%, the redundant triangular surface compression rate is improved by 11.9%, and the frame time standard deviation is reduced. Therefore, the method can dynamically and smoothly adjust the model details under different watching conditions, guarantees the consistency of visual effects, and remarkably improves the rendering efficiency. According to the method, an efficient and adaptive rendering solution is provided for the field of virtual reality.

The invention relates to the technical field of image processing, and discloses an AI image super-resolution reconstruction method based on a multi-scale fusion mechanism. The AI image super-resolution reconstruction method based on the multi-scale fusion mechanism comprises the following steps: establishing a binocular image system; establishing a double-end collaborative model architecture; establishing a front-end model optimization mechanism; according to the method, through a scene adaptive feature labeling decision model, three-dimensional position driven quantization and spectral response calibration replace manual presetting, and parallax-pose linkage correction is combined, so that subjective interference is thoroughly eliminated, and the misalign error suppression effect is improved by more than 40%; scale intelligent selection and cross-scale feature interaction are introduced into a dynamic interactive multi-scale attention convolutional neural network (DI-MSCNN), and dynamic definition of convolution kernel parameters is matched, so that feature extraction efficiency is improved by 30%-40%, and texture density differences can be accurately adapted; a detail hierarchical perception GAN (DLP-GAN) generates a strategy through hierarchical discrimination and detail partitioning.

Embodiments of the present application relate to a method and system for three-dimensional modeling of a train based on a binocular disparity prediction model, the method comprising: setting installation, acquisition and splicing rules of double parallel linear array cameras; setting a binocular disparity prediction model; acquiring first data set based on the installation, acquisition and splicing rules; training the binocular disparity prediction model based on the first data set; installing the double parallel linear array cameras after the training; when a train passes through the two installed cameras on the current train track, acquiring and splicing images to obtain images I1 and I2, inputting the images I1 and I2 into the binocular disparity prediction model for prediction to obtain a disparity map D 1‑2 , and based on I1, I2 and D 1‑2 , the present application can improve prediction accuracy in high-speed scenes and complex lighting environments, effectively overcome the limitations of a single perspective, and improve the integrity of three-dimensional reconstruction of edge and occluded areas.

A camera parameter automatic optimization system for binocular depth estimation, a true value acquisition module acquires a depth true value map in a simulation environment; an Auto-CAM learning module adopts a deep reinforcement learning framework based on an actor-critic, learns in reverse propagation according to a current camera configuration, a state vector generated by left and right view images and a depth error, so as to optimize camera parameters such as focal length, baseline distance and distortion coefficient; a deployment module outputs an optimal camera parameter configuration under a specific scene, a camera calibration module acquires accurate camera parameters through corner scanning, an image acquisition module acquires RGB binocular images, a depth map prediction module outputs a predicted disparity map by using a stereo matching network algorithm and converts the predicted disparity map into a depth map, and finally, a visual rendering module performs filtering and color processing. The application not only improves the accuracy of depth estimation and the adaptability of the system, but also improves the accuracy and efficiency of camera calibration, optimizes the user experience and visual effect.

The application discloses a binocular disparity map acquisition method, device and system for small targets, which is used for improving disparity optimization effect of small targets. The method comprises the following steps: performing hierarchical feature extraction on left and right views to obtain a feature map, compressing the feature map to 32 channels through a convolution operation with structure information, and constructing an initial splicing body; obtaining attention weights of corresponding stereo images of the left and right views, screening the initial splicing body by using the attention weights, constructing a cost volume according to a screening result and an initial disparity loss; fusing convolution context information and intermediate features after preliminary aggregation to aggregate the cost volume, and obtaining an aggregation result; supervising the aggregation result by using an adaptive multi-modal cross-entropy loss function, and performing multi-modal output on the supervised aggregation result by using a multi-modal disparity estimator, so as to obtain binocular disparity maps of the left and right views.

The present invention belongs to the field of image processing and computer vision, and relates to a fusion target recognition method based on UV disparity detection and YOLOv5. This method fully integrates the improved UV disparity detection and YOLOv5 model, uses the improved UV disparity detection to roughly identify pedestrians and vehicles and non-standard obstacles on the road, inputs the UV detection results into the YOLOv5 deep learning model, and fuses the two results for target recognition to obtain a highly stable and robust detection effect, and the recognition range is not limited to the type of target, thereby realizing the fusion target recognition function of binocular stereo vision. The present invention can efficiently and quickly realize the detection of drivable areas on the road, the detection of non-standard obstacles, and the classification and recognition of targets. It is a fusion target recognition algorithm with high stability and high robustness that combines the advantages of traditional binocular detection and deep learning target detection.

The present application relates to the field of video acquisition technology, and in particular to a beam, a binocular vision device and a drone thereof, wherein the beam comprises: a beam body, wherein a first mounting hole is respectively provided at both ends of the beam body; and an angle adjustment component, wherein the angle adjustment component comprises a piezoelectric stack, which is mounted on the beam body and is used to deform after being energized to drive the beam body to move, so as to adjust the angle between the axes of the two mounting holes so that the axes of the two mounting holes are parallel. During use of the beam of the present application, if the beam body is deformed, the piezoelectric stack of the angle adjustment component can drive the beam body to move, so as to adjust the angle between the axes of the two mounting holes so that the axes of the two mounting holes are parallel. In this way, the lenses respectively mounted on the mounting holes can always maintain the consistency of the axis angles when working, so as to avoid the degradation of the parallax map of the photosensitive component and ensure the sensing accuracy and normal use of the photosensitive component.