93 results about "Stereo matching" patented technology

The application provides a plate tail intelligent grading equipment and method based on visual guidance, and the core is that through integration of multi-view visual perception, adaptive grabbing, digital twin and cloud-edge collaborative control technology, full-process intelligent management of plate tail from warehousing, grading storage to on-demand use is realized; multi-view images of the tail are collected by a camera array, three-dimensional point clouds are reconstructed through a stereo matching algorithm, and morphological characteristic parameters of the tail are extracted; with the help of digital twin technology, a virtual model of each tail is generated and synchronized, a stacking strategy is optimized based on a genetic algorithm, and space utilization is maximized; when a use request is received, the system can perform virtual cutting pre-performance in the digital twin environment, accurately match the demand, and use the stacked tail through the temporary storage-backfilling mechanism; finally, through the cloud-edge collaborative platform, visual monitoring and global scheduling optimization of the inventory and equipment state are realized, and the tail management efficiency and material utilization are significantly improved.

A system for generating low-computational-resolution depth maps using a low-resolution camera is configured to: acquire a pair of stereo images; and generate a depth map by performing stereo matching on the pair of stereo images. The system is also configured to: acquire a first image including first texture information for the environment, the first image having a first image resolution higher than the image resolution of the stereo image pair. The system is further configured to: generate a reprojected first image by reprojecting the first image to correspond to an image capture viewpoint associated with the depth map. The reprojection of the first image is based on depth information from the depth map and includes the first texture information reprojected for the environment. The system is also configured to generate an upsampled depth map based on the depth map.

The application discloses a kind of coal mine underground image stereo matching methods based on threshold and weight Census transformation, comprising the following steps: S1, image information is collected by two monocular camera modules binocular holder;S2, the gray value of all pixels in support window is thresholded;S3, improved center point pixel calculation method obtains matching generation value;S4, improved dynamic cross-domain obtains cost aggregation value;S5, disparity value is obtained using WTA strategy, S6, the overall process is verified on the visual system of underground unmanned auxiliary transport vehicle above-mentioned.The application applies threshold and weight Census transformation method to coal mine underground perception, realizes the autonomous obstacle avoidance and visual reconnaissance function of coal mine underground unmanned auxiliary transport vehicle, reduces the influence of factors such as dust, unstable illumination conditions on stereo matching, improves the accuracy of stereo matching.

The application relates to a stereo matching and disparity map optimization method based on adaptive smooth constraint, which comprises the following steps: acquiring satellite stereo pairs and refined rational polynomial coefficients; processing the satellite stereo pairs and the rational polynomial coefficients by adopting a core line resampling method based on orthographic correction to generate core line image pairs; performing stereo matching on the core line images by adopting a semi-global matching algorithm based on adaptive parameters to generate an initial disparity map; performing optimization processing on the initial disparity map by using an improved weighted least square algorithm to obtain a final disparity map; and generating point clouds by spatial forward intersection based on the final disparity map, performing gridding processing on the point clouds, and acquiring a final digital surface model. Compared with the prior art, the application realizes the balance between noise suppression and detail reservation, thereby improving the accuracy and completeness of the DSM.

This application relates to a virtual binocular speckle stereo matching method based on local gray-level plane binary segmentation. It pertains to the fields of computer vision, 3D measurement, and stereo vision, and includes the following steps: S1: image loading and parameter settings; S2: local gray-level plane binary segmentation; S3: disparity calculation and sub-pixel optimization based on Hamming distance; S4: disparity map post-processing; sub-pixel interpolation is performed based on the matching cost curve to obtain disparity values ​​with sub-pixel accuracy, resulting in an initial sub-pixel disparity map; S5: depth map calculation and effective value filtering: the initial sub-pixel disparity map is filtered, consistency checked, and outlier removed to obtain an optimized dense disparity map; finally, combined with the calibration parameters of the virtual binocular system, the dense disparity map is converted into a depth map, and the result is visualized. This method has the advantages of high robustness, high accuracy and efficiency, and strong versatility.

The application discloses a PCB solder paste printing three-dimensional defect detection system based on multispectral imaging, and relates to the field of machine vision detection.The technical scheme points of the application include a multispectral image acquisition module, a four-channel industrial camera based on RGB and near-infrared, and a tunable LED light source array; the reflection characteristic data under different penetration depths are acquired through a wavelength switching mechanism; a motion control trigger module is based on a servo motor driven XYZ three-axis object table, and positioning is realized in cooperation with an encoder feedback; a pulse width modulation signal is adopted to coordinate the camera shutter and the platform moving speed; the application realizes the dual verification of material composition quantitative analysis and three-dimensional topography reconstruction through the combination of the four-channel imaging of RGB and NIR and the Beer-Lambert law modeling, and avoids the limitations of monocular vision; the cross-frame data alignment is realized through the projection of a structured light and a binocular stereo matching technology in cooperation with an ICP algorithm, and the details defects can be better detected.

A virtual stereovision imaging array system and a moving target three-dimensional reconstruction method, the system comprises multiple cameras and multiple galvanometer mirrors, the galvanometer mirrors and the cameras are matched one by one, and a real-time controllable virtual camera array is formed; through synchronous control of the galvanometer mirrors, each virtual camera simultaneously tracks and locks the moving target; based on the real-time pose relationship between the virtual cameras and combined with the synchronously acquired image information, real-time three-dimensional reconstruction of the moving target is completed. The reconstruction method comprises joint calibration of the matched galvanometer mirrors and the cameras, determination of camera internal parameters and virtual camera external parameters; through real-time synchronous adjustment of the galvanometer mirrors, each virtual camera cooperatively tracks the moving target, acquires the moving target image, and updates the virtual camera external parameters in real time; in each frame of image, stereo matching is performed, and combined with the real-time updated virtual camera external parameters, the three-dimensional point cloud of the moving target is solved and obtained. The present application can complete high-efficiency and high-precision three-dimensional reconstruction of a high-speed moving target in a large field of view.

This application discloses a method and system for vehicle scene depth estimation based on multi-scale attention and lightweight 3D convolution, relating to the field of stereo matching. The method includes: acquiring a binocular view; extracting multi-scale features using a lightweight feature extraction network; extracting multi-scale contextual features using a multi-scale spatial attention network; constructing a 4D cost volume on the 1 / 4 scale features using a group correlation method, and regularizing it through a lightweight 3D convolutional network; performing disparity regression on the aggregated 4D cost volume to obtain an initial disparity map. Based on the multi-scale contextual features and the initial disparity map, a hierarchical residual refinement module based on convolutional gated recurrent units is used for multi-round residual learning and iterative updates to output the scene disparity estimation result. This application achieves high-precision recognition of the depth of field in front of the vehicle while ensuring computational efficiency and real-time performance, balancing the algorithm's prediction speed and accuracy, and improving the safety of vehicles driving in complex environments.

The application belongs to the technical field of medical treatment and can be used for sample data enhancement in the scene of human body, tissue organ modeling and the like in the medical field, and particularly relates to an image data enhancement method, a training method and device of a binocular stereo matching model. The method comprises the following steps: obtaining a target sample image pair, the target sample image pair comprising a first image and a second image; generating a first random mask and a second random mask; obtaining a first disordered image and a second disordered image; obtaining a first data enhancement image based on the first random mask, the second disordered image and the first image; obtaining a second data enhancement image based on the second random mask, the first disordered image and the second image; and combining the first data enhancement image and the second data enhancement image to obtain a data enhancement sample image pair. The above method and device improve the adaptability of the introduced noise to the training sample set, so that the expanded training sample has better training effect.

This invention discloses a visibility estimation method and apparatus based on binocular vision. The method includes: simultaneously acquiring left and right views using a calibrated binocular camera; inputting the image pairs into a stereo matching deep neural network to obtain a disparity map and converting it into a depth map; inputting the left view and depth map into a dual-branch deep convolutional neural network to extract multi-scale features; fusing RGB features and depth features through a cross-modal feature fusion module; and finally outputting a visibility estimate through a feature aggregation and regression module. The apparatus includes a binocular image acquisition unit, a data processing and visibility estimation unit, and a result output unit. This invention solves the depth ambiguity problem of monocular vision by fusing binocular depth information and image appearance information, achieving high-precision, non-contact visibility surface measurement. It has the advantages of low cost and flexible deployment, and is suitable for visibility monitoring in traffic scenarios such as highways.

The application discloses a night highway pit and pond recognition method and system based on low-illumination enhancement and parallax calculation, and relates to the technical field of intelligent traffic monitoring and road maintenance. The method comprises the following steps: S1: collecting environmental data, calculating a meteorological index to determine a laser compensation power, and obtaining a stereo image pair to generate environmental metadata; S2: determining an enhancement strategy based on the environmental metadata, and obtaining an enhanced image pair by using an improved Retinex algorithm; S3: performing stereo matching on the enhanced image pair, and generating a scene depth map by fusing pose data; S4: based on a double-branch recognition model, fusing image and depth features, detecting and verifying a pit and pond area, and obtaining a pit and pond area mask; and S5: based on the pit and pond area mask, the depth map and the pose data, calculating three-dimensional parameters and a position of the pit and pond, and generating a detection report. Through adaptive image enhancement and multi-source data fusion technology, the accuracy of pit and pond recognition and the three-dimensional measurement precision in a night low-illumination environment are improved, and automatic and efficient inspection without affecting traffic is realized.

The application discloses a stereo matching method, device and equipment and a computer readable storage medium. The hidden state is decoupled from the update matrix of the disparity map through a multi-scale decoupling LSTM network, more high-frequency information is retained, and further based on normalized refinement processing, information from the up-sampled disparity map, i.e. the original left and right images containing high-frequency information, is fully utilized to enhance edges and details, solving the problem that in the original GRU structure, the information of the update matrix for generating the disparity map is coupled with the hidden state transition value between iterations, so that high-frequency details in the hidden state are difficult to retain. Meanwhile, in order to balance the performance and the calculation speed, the resolution of the iteration stage is at most 1 / 4 of the original resolution, so that it is difficult to produce a disparity map with sharp edges and subtle details.

Systems and methods are disclosed that use a Foundational Stereo Model to generate an output disparity map. The Foundational Stereo Model includes side-tuning adapters (STA) that utilize a vision transformer (ViT) and a convolutional neural network (CNN) to generate feature maps. Specifically, the CNN may be used to adapt the ViT-based monocular depth estimation network for the stereo setup, which synergizes the strengths of both CNN and ViT architectures. In addition, the Foundational Stereo Model includes an attentive hybrid cost filtering (AHCF) that uses two branches that also utilizes the advantages of both a transformer architecture and the CNN architecture. Furthermore, the Foundational Stereo Model may perform iterative refinement of an initial disparity map to obtain the output disparity map based on performing a convolutional gated recurrent unit (GRU) operation.

This invention discloses a pixel-level visibility and fog structure joint modeling method and system based on binocular depth estimation and pixel-by-pixel transmittance fusion. The method utilizes binocular vision to obtain accurate scene depth and constructs a two-stage neural network: the first stage estimates depth through stereo matching; the second stage fuses depth and fog map texture features, jointly regresses pixel-by-pixel transmittance and visibility maps, and detects localized patchy fog by analyzing their spatial distribution. This invention overcomes the limitations of monocular methods due to the lack of geometric information, achieving physically interpretable high-precision visibility perception and effectively identifying non-uniform fog conditions that endanger traffic safety. The system can run in real-time on an embedded platform, combined with a gimbal for panoramic monitoring, and is applicable to fields such as intelligent transportation, autonomous driving, and meteorological observation.

The present application relates to machine vision, in particular to a kind of dense feather layering grasping method based on from coarse to fine stereo vision, utilize binocular camera synchronous acquisition RGB image pair, after input first stereo matching neural network model after down-sampling, obtain global rough depth map;After input instance segmentation neural network after RGB image and global rough depth map are fused, obtain the segmentation mask and ordering score of each identified upper layer feather instance;According to the segmentation mask of target feather instance, the region of interest is cut out from RGB image pair, after up-sampling processing, input second stereo matching neural network model, obtain the fine three-dimensional point cloud of target feather instance;The optimal grasping posture of manipulator end effector is calculated, and the grasping point coordinates are converted to manipulator base coordinate system according to transformation relationship, control manipulator to complete grasping operation;The present application can overcome the defect that it is difficult to accurately visually grasp the non-rigid object such as dense stacked feather.

This application provides a stereo matching ranging method and system for dynamic occlusion scenes. The method includes: acquiring an image dataset and a point cloud dataset of a ranging area over a preset time period, and determining target image data; performing optical flow calculation on the target image data based on the image dataset to generate motion information of each pixel in the target image data; inputting the target image data into a preset convolutional neural network to determine several dynamic object regions; generating tilt support windows for each pixel based on the motion information of each pixel and each dynamic object region using a preset stereo matching algorithm, and then generating a disparity map corresponding to the target image data based on each tilt support window; and generating a corresponding complete depth map based on the disparity map, the image dataset, and the point cloud dataset using occlusion repair technology to determine the distance measurement value from each pixel to the subject being photographed.

The present application relates to remote sensing mapping and computer vision technical field, especially to the hilly and mountainous area cultivated land monitoring protection method fusing low altitude AI perception technology, including the following steps: S1, the original aerial image sequence with preset overlap rate and attitude position information of the monitoring area are acquired;S2, based on the original aerial image sequence and attitude position information, a three-dimensional dense point cloud is constructed by using a multi-view stereo matching algorithm, and the three-dimensional dense point cloud is subjected to visibility analysis, the texture mapping of the projection occlusion area is removed, and a true projection image is generated, S3, the three-dimensional dense point cloud is subjected to cloth simulation filtering, in the present application, the ground point extraction is carried out on the three-dimensional point cloud, and a digital terrain model is generated, the real ground slope value is calculated, the slope value is combined with the crop type of the land plot to realize the objective quantitative monitoring of the hilly and mountainous area steep slope cultivation behavior.

This invention relates to the field of vehicle assisted driving technology, and discloses a method, system, and electronic device for three-dimensional matching of wheel trajectories. The method includes: acquiring a first image and a second image with a fixed spatial positional relationship, which are synchronously acquired by an image acquisition device; acquiring a current frame image from the first image or the second image; estimating an initial wheel trajectory based on the vanishing point in the current frame image and the vehicle's inherent parameters; iteratively updating the trajectory parameters by combining real-time steering wheel angle and vehicle speed; determining the region of interest and performing feature extraction and matching on the image within the region; optimizing through cost aggregation and extracting a single disparity value along the trajectory path. This method solves the problems of high computational cost, difficult deployment, and difficulty in adapting to dynamic changes in vehicles with fixed regions of interest in large-scale deep learning networks, thereby improving the accuracy and real-time performance of wheel trajectory tracking and providing a more reliable perception basis for trajectory prediction and active control of vehicle assisted driving systems.

The application provides a point cloud completion method and device and a storage medium, which comprises the following steps: collecting original images and stripe images of an object through a binocular structured light system; reconstructing a point cloud with high precision but with reflection and dark area missing by using the stripe images; meanwhile, calculating a disparity map by using a FoundationStereo deep learning model for stereo matching based on the original images to obtain a point cloud which is complete but has lower precision than the stripe image reconstruction; in order to meet the point cloud coordinates in the two three-dimensional reconstruction methods, the point cloud coordinates of the two reconstruction methods are corrected based on a standard ball multi-position reconstruction data calculation error model; firstly, the two point clouds are spatially registered and aligned; then, the missing area of the high-precision point cloud is analyzed, the corresponding area in the low-precision point cloud is located through image mapping, and the point cloud of the corresponding area is extracted as a patch. The application effectively solves the data missing problem of the structured light measurement in the reflection and dark area, and realizes the balance between the point cloud precision and completeness.

The application discloses a wave image matching method based on fusion matching cost, and relates to the field of stereo matching, and comprises the following steps: S1, left and right eye images of waves are collected by using binocular cameras to determine a disparity range; S2, the left and right eye images of original waves collected in the step S1 are preprocessed; S3, a matching cost value of the wave images is obtained by using an improved Census algorithm; S4, improved Census cost, AD cost and gradient cost are fused to obtain a cost space; S5, a cross-domain is used for cost aggregation; S6, a winner-takes-all (WTA) strategy is used to calculate the disparity of the wave images; S7, the disparity map in the step S6 is subjected to left-right consistency checking, and is subjected to hole filling and sub-pixel optimization processing. The application adopts the above method, solves the problem that the disparity map obtained from the wave images is prone to a large number of invalid points and mismatching points, improves the correctness of wave stereo matching, and guarantees the matching precision of the depth discontinuous region.

This invention discloses a pixel-by-pixel cross-medium 3D measurement method based on a ray model, belonging to the field of optical measurement technology. Addressing the technical challenge of imaging distortion caused by light refraction of the object under transparent media coverage, this method proposes a novel calibration and reconstruction framework that does not rely on precise modeling of medium parameters. By constructing a structured light projection system and utilizing imaging data of a calibration target in multiple height planes, the method independently calibrates the corresponding object-side ray equation for each camera pixel. Combined with binocular stereo matching and pixel-height mapping techniques, it ultimately achieves high-precision 3D surface reconstruction of cross-medium scenes. This invention is applicable to optical measurement scenarios with unknown or complex refractive interfaces, aiming to overcome the bottlenecks of difficult parameter solving and limited accuracy of traditional pinhole models in cross-medium environments, providing an innovative solution for 3D measurement in complex environments such as underwater and inside protective enclosures.

The application discloses a kind of based on infrared binocular vision's offshore target detection and three-dimensional point cloud perception method, the method includes by constructing based on EMA attention module and DCNv2 module improvement YOLOv8 network, the YOLOv8-IRECG target detection model obtained;Through sample data set, the optimal target detection model is obtained by model training to YOLOv8-IRECG target detection model, to realize the category detection of offshore target and output two-dimensional detection frame;Based on improved SGBM stereo matching algorithm, the depth feature map is obtained by stereo matching to preprocessed left image and preprocessed right image;Based on the camera internal parameter matrix of prelabeling, according to the depth feature map and the two-dimensional detection frame output by optimal target detection model, the offshore target detection and three-dimensional point cloud perception based on infrared binocular vision are realized.The application solves the problem that existing method cannot realize high-precision target detection and three-dimensional positioning in low-visibility offshore environment, and outputs the comprehensive solution of rich three-dimensional point cloud data.

The application discloses a kind of based on laser feature point's multi-view stereo vision steel plate flatness detection method and system, belong to the field of steel plate flatness detection, comprising: by projecting parallel laser stripe to steel plate surface, at least two groups of binocular vision module with overlapping field of view are used to synchronous acquisition image;Image processing is extracted laser stripe subpixel center coordinates, by stereo matching and three-dimensional reconstruction obtains local three-dimensional point cloud;Further, each point cloud is spliced and fused into complete three-dimensional point cloud covering the whole surface of steel plate;Finally, through plane fitting and gridding distance calculation, the quantitative evaluation of steel plate overall and local flatness is realized.The system can realize non-contact, high efficiency, high precision online automatic detection, effectively overcome the problems of low efficiency, incomplete data, poor environmental adaptability of traditional contact measurement, suitable for steel plate quality automatic detection and digital management in continuous production line.

Embodiments of the present application provide an asymmetric binocular industrial endoscope depth of field extension method, device, equipment and medium, relate to the industrial nondestructive testing technical field, the method comprises: by obtaining the target close focus image and target far focus image after correction, then the target close focus image and target far focus image are stereoscopic matching, obtain corresponding first parallax diagram, and the first parallax diagram is converted into corresponding physical depth diagram, and the fusion weight mask film corresponding to the physical depth diagram is generated, then according to the first parallax diagram, the pixel point re-projection is carried out to target far focus image, the view angle of target far focus image is transformed to the view angle of target close focus image, obtain the synthetic far focus image that is aligned with target close focus image in space, finally, according to the fusion weight mask film, target close focus image and synthetic far focus image are weighted fusion, obtain corresponding panoramic depth image.

This invention discloses a method and apparatus for determining the tilt of a power transmission tower, belonging to the field of computer vision technology. The method includes: inputting first image data into a key point detection model and outputting the pixel coordinates of key points of the power transmission tower; performing stereo matching in second image data using the pixel coordinates of the tower base and the center of the tower top to determine the relative three-dimensional coordinates of the tower base and the center of the tower top in the camera coordinate system; performing coordinate transformation on the relative three-dimensional coordinates based on the attitude and position data of a UAV to generate the absolute three-dimensional coordinates of the tower base and the center of the tower top in the world coordinate system; determining the base reference center coordinates of the power transmission tower based on the absolute three-dimensional coordinates of the tower base, and calculating the horizontal projection offset and vertical height difference based on the absolute three-dimensional coordinates of the center of the tower top to determine the tilt of the power transmission tower. By implementing this invention, the problem of inaccurate determination of the tilt of power transmission towers in the prior art can be solved.

The application provides a face recognition and living body detection fusion method and device based on a binocular camera, and the method comprises the following steps: after collecting original left and right face images, performing binocular image preprocessing to obtain a standard binocular image pair; performing multi-scale stereo feature extraction on the standard binocular image pair to obtain a depth-texture joint feature representation and a disparity map; combining the disparity map, performing adaptive 3D face reconstruction and time sequence dynamic feature modeling, and through multi-level living body detection fusion decision and face feature extraction and matching, completing end-to-end fusion verification of face living body detection and face recognition. Through deep fusion of binocular stereo vision and deep learning, multi-dimensional living body discrimination features are constructed, and various known and unknown attacks can be effectively resisted. An adaptive stereo matching and 3D face reconstruction mechanism is designed, the unstable recognition problem caused by illumination change and posture change is overcome, and the system robustness is improved.

This invention relates to an image acquisition method and related equipment for a structured light 3D scanning system, encompassing the fields of machine vision and 3D measurement technology. The main control unit configures the physical projection range of the projection module as the data reading window of the camera hardware. Data processing is performed on the smallest rectangle containing two correction masks, allowing the camera to physically read and transmit only pixel data containing valid encoded information. This reduces data transmission volume, shortens single-frame transmission time, and increases the system's acquisition frame rate. Simultaneously, since the algorithm's input area is verified by stereo correction constraints, data processing naturally satisfies the row alignment conditions required for stereo matching, eliminating the need for secondary correction at the algorithm level and further saving computational resources. This achieves the technical effect of increasing the transmission frame rate by utilizing the camera hardware bandwidth and compressing the time spent on data transmission in the scanning system, enabling the system to achieve high-frame-rate real-time acquisition while ensuring 3D measurement accuracy.

The application provides a three-dimensional scanning system, a three-dimensional scanning method and a computer readable storage medium. The system comprises a transmitting end, a receiving end and a processor. The transmitting end transmits a composite line pattern light beam comprising a dense line pattern and a sparse line pattern to a scanned object. Each coded line in the sparse line pattern is at least aligned with part of the measurement lines in the dense line pattern to achieve unique coding. The receiving end collects the composite line pattern light beam reflected by the scanned object and generates left and right composite line images. The left and right composite line images each comprise a dense line image and a sparse line image. The processor decodes the dense line image in the left and right composite line images according to the arrangement number of each coded line in the sparse line image to identify a plurality of measurement lines. The depth information of the scanned object is obtained by using the measurement lines in the identified left and right dense line images for stereo matching. The system provided by the application can obtain high-precision three-dimensional scanning information while reducing the cost.

The application relates to a stereo matching implementation method and device for real-time depth estimation of a mobile device, which comprises the following steps: performing matching cost calculation on a target image pair to obtain a cost matrix corresponding to a main image in the target image pair; based on the cost matrix, adopting a reverse path synchronous aggregation mode to determine the aggregation cost of each pixel in the main image under each disparity; according to the aggregation cost, adopting an expanded thread data slice mode to perform disparity calculation on each pixel in the main image; and performing disparity optimization on each pixel in the main image to obtain a disparity map corresponding to the main image. In the cost fusion calculation, the cost aggregation is simultaneously performed on two paths which are reverse paths of each other, so that the access cost of the global memory is reduced. Meanwhile, in the disparity calculation, the number of concurrent threads is reduced through the expanded thread data slice mode, and then the thread synchronization cost is reduced.