91 results about "Corner point detection" patented technology

The invention relates to the technical field of image stitching, and discloses a video transmission high-definition image intelligent stitching method and system. The system comprises an image preprocessing module, a feature point matching and screening module, an image optimal splicing seam generation module and a video output module. The method comprises the following steps: firstly, acquiring a video, carrying out dynamic range equalization, carrying out filtering processing by using an adaptive noise reduction method, and carrying out image correction; secondly, extracting feature points by using a self-adaptive corner detection algorithm, matching the feature points based on a multi-dimensional spatial data index to obtain feature point matching pairs, and screening the feature points; searching an overlapping region, and introducing a dynamic search algorithm to generate an optimal image splicing path to obtain an optimal image splicing seam; and finally, dividing the image according to the dynamic grid, and realizing video output by using priority ranking. According to the method, the video images are processed and spliced, the purpose of intelligent image splicing is achieved, and the method is accurate and objective.

The invention discloses a deformation monitoring method and system based on a machine vision technology, and relates to the technical field of crossing of structural health monitoring and computer vision, and the method comprises the steps: improving the angular point positioning precision in a calibration stage through a sub-pixel-level Harris angular point detection algorithm; robust recognition and initial positioning of an artificial marker in a first frame of image are realized by using an X-Feat convolutional neural network model, an anti-interference template is generated in combination with histogram equalization and Gaussian filtering, the stability of the system is enhanced, a local search area is delimited by taking an initial positioning point as a center, and a deep learning and interpolation algorithm is combined, so that the anti-interference performance of the system is improved. High-precision tracking and displacement calculation of the marker center are realized, the monitoring resolution is effectively improved, pixel-level displacement is accurately converted into physical space displacement through inverse transformation of a homography matrix, and dual threshold criteria of mode length and change rate are introduced for dynamic early warning, so that the engineering practicability and safety response capability of a monitoring result are improved.

The invention provides a pallet position identification system and method, and relates to the technical field of image identification, and the method comprises the steps: carrying out the Gaussian filtering and Canny edge detection of image data, extracting the edge features of pallets, judging whether there is an overlapped pallet in combination with depth information, and carrying out the layered identification of an overlapped region to mark a target pallet. Calculating a contour vector of the target pallet, and mapping the contour vector to a three-dimensional space by using depth information to obtain a spatial position coordinate; when the position meets a safe distance threshold value, angular point detection is carried out, coordinates of four angular points are extracted, orientation angles are calculated, and complete position features are output. According to the method, the problems of incomplete pallet edge detection, large space positioning error and inaccurate posture recognition are effectively solved, the pallet detection precision and reliability are improved, and meanwhile, the carrying safety in a storage environment is ensured.

The invention discloses a port container loading and unloading positioning system and method based on visual identification, and relates to the technical field of port logistics automation. The method comprises the following steps: acquiring a container surface image through an image acquisition device on a sling, and detecting an initial angular point through a deep learning model; when the angular points are missing or the confidence coefficient is insufficient, edge features are extracted, and sub-pixel coordinates are predicted through a reconstruction network so as to complement the angular points; sorting the complete angular point set to form closed-loop vectors, and calculating cross products of adjacent vectors to verify the topological effectiveness; and when the topology is effective, through coordinate conversion and pose optimization, outputting three-dimensional coordinates of the angular points in a world coordinate system, and further determining the position of the container and feeding back the position to a lifting appliance control system to guide loading and unloading. The system correspondingly comprises an image acquisition and corner detection module, a corner completion module, a corner sorting and cross product calculation module, a topology verification and coordinate conversion module and a positioning feedback module. The problem of inaccurate positioning caused by lifting tool shaking, stacking deviation and the like is solved, and the precision and efficiency of container loading and unloading are improved.

The invention discloses a target vehicle attitude recognition method based on 3D point cloud feature extraction, and the method comprises the steps: carrying out the preprocessing of the point cloud data of a target vehicle obtained through a laser radar, and the preprocessing comprises Gaussian statistical filtering, voxelization sampling, and vehicle tire point cloud extraction based on a region growing algorithm; based on the point cloud data, key points of tire steering features are extracted, and a Harris corner detection method is used for detecting the key points; calculating a fast point feature histogram (FPFH) descriptor based on the extracted key points, and performing coarse registration by adopting an SAC-IA algorithm to obtain a coarse registration matrix; and using a point-to-surface nearest point iteration ICP algorithm to perform fine registration on the point cloud after coarse registration to obtain attitude information of the target vehicle. According to the method, the three-dimensional attitude angle of the target vehicle is accurately estimated by preprocessing the laser radar point cloud, extracting tire key points and carrying out SAC-IA coarse registration and ICP fine registration. Therefore, vehicle attitude recognition with high shielding robustness and high precision is realized, and the safety of automatic driving is enhanced.

The invention provides a 4D millimeter-wave radar and camera calibration method based on projection errors. The method comprises the following steps: acquiring radar point cloud and an original image of a calibration board in a static scene through a 4D millimeter-wave radar and a camera; performing dynamic object filtering and noise suppression on the synchronous data to obtain a static enhanced point cloud and a corrected image; performing coordinate extraction and checkerboard corner detection on the filtered data to obtain point cloud 3D coordinates and pixel 2D coordinates; performing projection conversion on the point cloud 3D coordinate through the external parameter matrix to obtain a conversion coordinate, and calculating a projection error between the conversion coordinate and the pixel 2D coordinate; and performing iterative optimization on the extrinsic parameter matrix through a nonlinear optimization algorithm based on the projection error to obtain an optimized matrix. According to the method, through dynamic interference suppression, noise collaborative filtering and a nonlinear optimization mechanism, the calibration precision and the environmental adaptability are remarkably improved.

The application discloses a camera calibration method, a camera calibration system and a computer readable storage medium, and relates to the field of image processing. The calibration method comprises the following steps: acquiring first image information, wherein the first image information is used for representing an image after coarse screening of connected domains of a preset calibration template image; performing connected domain screening according to an overlap ratio and an overlapping area of a minimum circumscribed rectangle and a positive circumscribed rectangle of each first reserved connected domain in the first image information, to obtain a plurality of second reserved connected domains; classifying and sorting the second reserved connected domains; determining corner point information according to a classification and sorting result; and obtaining a conversion relationship between a pixel coordinate system and a world coordinate system according to the corner point information. The camera calibration method improves the accuracy and efficiency of corner point detection through twice different connected domain screening, and optimizes the effect of camera calibration.

The present application relates to the technical field of visual detection, and discloses a kind of offshore floating photovoltaic power station cleaning robot edge distance calculation method, comprising the following steps: S1, image and radar azimuth acquisition;S2, image correction and pretreatment;S2-1, based on the aforementioned direction angle, image is sequentially subjected to histogram equalization, rotation and projection processing, and the image horizontal line is 0 ° image;S3, edge and corner point detection are carried out by image recognition system, and point set is obtained;S4, point set is used to remove impurities algorithm, and block straight line fitting is carried out, and intercept data is obtained;S5, in combination with camera inclination, height and straight line intercept data, the actual cleaning robot edge distance is calculated.By in the process of robot normal work, distance photovoltaic panel edge angle and distance are judged at any time, to avoid cleaning robot falling into the sea, to improve the cleaning rate of photovoltaic power station and the safety factor of itself.

The embodiment of the invention provides a parking space detection method and device and a vehicle, and the method comprises the steps: obtaining original images shot by a plurality of camera devices on the vehicle, the original images being used for representing an image obtained by shooting at least one parking space around the vehicle, and the different original images having different shooting angles of view; splicing the original images based on camera shooting parameters of camera shooting equipment to obtain a target image; the parking space is detected on the basis of the original image and the target image, parking space detection results are obtained, and the parking space detection results at least comprise a target corner point detection result corresponding to the corner points of the parking space and a target entrance detection result corresponding to the entrance of the parking space. According to the invention, the technical problem of low accuracy of parking space detection in the prior art is solved.

The application relates to the technical field of image recognition, and particularly provides an unmanned aerial vehicle part automatic recognition method based on corner point detection, which comprises the following steps: constructing a neural network model comprising a channel attention mechanism, acquiring multiple unmanned aerial vehicle images, and pre-processing and unmanned aerial vehicle part labeling the unmanned aerial vehicle images; acquiring corner point information in the unmanned aerial vehicle images after the unmanned aerial vehicle parts are labeled through a corner point detection method, and generating a corner point feature map; fusing the corner point feature map and the unmanned aerial vehicle images after the unmanned aerial vehicle parts are labeled to obtain a corner point image; and training the neural network model again by using the corner point image, inputting a to-be-detected target image into the trained neural network model, and recognizing unmanned aerial vehicle parts in the to-be-detected target image. The application enhances the robustness of the neural network model to complex scenes and noise interference, and realizes accurate and efficient unmanned aerial vehicle part automatic recognition.

The application provides an AVM calibration method, comprising the following steps: determining at least one original calibration pattern in which corner point detection fails in a plurality of original calibration patterns laid around a vehicle and an original calibration pattern arrangement in which the at least one original calibration pattern is located; acquiring a redundant calibration pattern arrangement arranged at intervals with the original calibration pattern arrangement; performing a corner point detection algorithm on the original calibration pattern arrangement and the redundant calibration pattern arrangement to obtain corner points of a calibration pattern arrangement in which corner point detection succeeds in the original calibration pattern arrangement and the redundant calibration pattern arrangement, so as to calibrate AVM by using the corner points. The application also discloses a vehicle-mounted device and a storage medium. The application realizes the following advantages: without repeatedly adjusting light, without a special material calibration pattern and without constructing a darkroom and arranging controllable light sources, the operation process of AVM calibration is simple.

The application provides a target positioning method based on panoramic segmentation, and belongs to the field of target positioning. Firstly, an image acquisition sensor is used to acquire a real scene image of a target area, and an initial pose of the sensor is acquired; a rendering image is generated according to the initial pose of the sensor and a three-dimensional geographical space model which has been established; then, panoramic segmentation is performed on the real scene image and the rendering image, corner point detection and matching are performed on the segmentation results, and matched point pairs of the two images are obtained; the real pose of the sensor is calculated by using the matched point pairs and the initial pose of the sensor; finally, the real pose of the sensor is used to realize target positioning in a real scene. Compared with the matching method in the prior art which uses a deep learning method, the application does not need a large number of samples, uses a corner point detection method to determine the matched point pairs of the two images, improves the work efficiency, improves the matching precision between the heterogeneous images, and uses the real pose of the sensor to realize target positioning, thereby improving the target positioning precision.

The application relates to a camera calibration method, device, equipment and computer readable storage medium. The method comprises the following steps: acquiring a first checkerboard image collected by a to-be-calibrated camera for a first checkerboard and a second checkerboard image collected by the to-be-calibrated camera for a second checkerboard, performing corner point detection on the first checkerboard image to obtain first corner points, and performing corner point detection on the second checkerboard image to obtain second corner points; determining an initial calibration model according to first image coordinates and first physical coordinates of the first corner points; determining candidate physical coordinates corresponding to the second corner points according to second image coordinates of the second corner points and the initial calibration model; aligning the second physical coordinates of the second corner points with the candidate physical coordinates to obtain target physical coordinates corresponding to the second physical coordinates; and determining a target calibration model according to the target physical coordinates and the second image coordinates. According to the application, high-precision calibration of a camera can be realized.

The application discloses a method for locating the wing point on the body surface based on NCCT image data and belongs to the technical field of medical image processing. The method uses the NCCT sequence data of a routine image inspection at any time to reconstruct the shape of a skull, automatically and individually identifies the eyebrow bone features through corner point detection and Kmeans mean clustering, accurately determines the outer edge points of the eyebrow bone, adopts the method that the outer edge of the eyebrow bone is pushed forward to the direction of the extension line of the eyebrow bone by a curved surface distance of 3.5 cm and the midpoint of the malar arch is pushed upward by a linear distance of 4 cm, and the wing point is automatically and quickly located, which has important clinical significance.

The application discloses an image segmentation method and device based on optical detection. The method comprises the following steps: acquiring a reference segmentation region of a target image; intercepting at least one sub-detection region in the reference segmentation region according to the boundary of the reference segmentation region, wherein the sub-detection region covers at least an intersection position of adjacent boundaries of the target detection region; performing corner point detection in the sub-detection region, and determining a target corner point according to the relative position of the intersection position of the adjacent boundaries in the reference detection region; and dividing a target segmentation region according to the target corner point. The technical scheme provided by the application improves the test efficiency and reduces the recognition error.

The present specification relates to the technical field of equipment inspection, and provides an equipment inspection method and device based on computer vision, and equipment, the method comprising: acquiring a target image of a target equipment; performing common corner point detection on the target image to obtain a common corner point coordinate set in the target image; determining whether the target image meets a preset condition according to the common corner point coordinate set; when the target image meets the preset condition, performing perspective transformation on the target image according to a pre-constructed perspective transformation matrix and the common corner point coordinate set to obtain a corrected target image; identifying an interest region in the corrected target image; the interest region contains an indicator for indicating a running state of the target equipment by color; performing similarity matching of a color of the interest region with a preset color, and determining an inspection result of the target equipment according to a matching result. The present specification can improve the inspection efficiency and application range of equipment inspection.

The present application belongs to the technical field of camera calibration, and discloses a focusing light field camera geometric parameter calibration method, which comprises the following steps: using a focusing light field camera to collect and acquire a white image and a corresponding checkerboard original light field image under different object distances; determining the focal plane of the camera based on the original light field image; using an edge diffusion function information fitting-based corner point detection algorithm to calculate the checkerboard corner point coordinate information in the original light field image on the focal plane; calculating the center coordinate information of each microlens in the microlens array of the camera and the size information of a single microlens in the microlens array based on the white image; calculating the pixel block size in the microlens sub-image taken by the center sub-aperture image according to the above information; and calculating the distance from the virtual imaging surface to the microlens array and the distance from the microlens array to the plane where the photosensitive element is located based on the pixel block size and in combination with a Gaussian imaging formula. The present application can improve the calibration accuracy of the geometric parameters of the focusing light field camera.

The application relates to the technical field of image splicing, and discloses a video transmission high-definition image intelligent splicing method and system; the system comprises an image preprocessing module, a feature point matching and screening module, an image optimal splicing seam generation module and a video output module; first, a video is acquired and dynamic range equalization is carried out, an adaptive noise reduction method is used for filter processing, and image correction is implemented; second, an adaptive corner point detection algorithm is used for feature point extraction, feature points are matched based on multidimensional space data indexing, a feature point matching pair is obtained, and feature point screening is carried out; then, an overlapping area is searched, a dynamic search algorithm is introduced to generate an image optimal splicing path, and an image optimal splicing seam is obtained; finally, the image is divided according to a dynamic grid, and video output is realized by using priority sorting; the application realizes the purpose of image intelligent splicing by processing and splicing the video image, and the method is accurate and objective.

This application discloses a method, circuit, device, electronic device, and storage medium for detecting image corner points, belonging to the field of electronic device technology. The image corner point detection method includes: processing the image to be detected to obtain N feature values ​​corresponding to N first pixels in the image, where N is a positive integer; processing the N feature values ​​to obtain distribution information corresponding to the N feature values; determining a corner point threshold based on the distribution information and the number of targets, where the number of targets is the number of target corner points; and determining the target corner points among the N first pixels based on the corner point threshold.

The embodiment of the application provides a rear sliding machine head conveying belt area identification method and device, the method comprises the following steps: through improving frame difference method, the rear sliding machine head image obtained is identified, and a motion area is obtained; according to the motion area and the pre-set feature selection size, corner point detection and clustering are carried out on the rear sliding machine head image, and a plurality of categories of feature subgraphs are obtained; through the plurality of categories of feature subgraphs and the motion area, the rear sliding machine head image is matched and identified, and a conveying belt area is obtained; the conveying belt area is obtained by improving the frame difference method and the feature matching method for the area matching identification of the rear sliding machine head image, all related feature pixel points can be completely extracted, the internal hole of the object is filled through the multi-frame accumulation effect, the moving object is accurately detected, so that the identification accuracy of the conveying belt area is improved, and the feature matching method can avoid the motion area detection failure caused by the stop of the conveying belt area.

The invention discloses a correction plate image corner detection method and an optical axis center correction method of a photoelectric pod, and the corner detection method comprises the steps: carrying out the preprocessing of a correction plate image, and obtaining a preprocessed image; detecting the preprocessed image by adopting a plurality of detection algorithms to obtain a plurality of detection images; fusing the plurality of detection images by adopting a neighbor aggregation and voting integration mode to obtain a fused candidate point set; rC grid sorting is carried out on the fusion candidate point set, and a temporary RC grid is obtained; and carrying out geometric refinement on the temporary RC grid, and outputting an angular point RC grid. According to the method, multiple detection methods are adopted for detection to obtain the detection images, fusion is carried out in a neighbor aggregation and voting integration mode, then through RC grid sorting and geometric refinement, the angular point RC grids of the integrated correction plate image are output, and the detection result is more stable than a single detection algorithm.

The present application relates to the technical field of image processing, and more particularly to a shaft part forging forming monitoring method and system. The method comprises: obtaining feature values of a pixel point in the surface image of a formed shaft part in multiple channels in HSV color space, and a gradient direction angle of the pixel point; determining local consistency of the pixel point; determining a weighted Euclidean distance between the pixel points; clustering all pixel points according to the weighted Euclidean distance to obtain a plurality of clustering clusters; determining gradient saliency and candidate corner point number of an image region where the clustering cluster is located; and obtaining candidate corner points by using a Harris corner point detection algorithm based on the candidate corner point number, and measuring whether the shaft part forging is qualified according to the Euclidean distance between the candidate corner points. The present application solves the problem of inaccurate recognition caused by large feature differences in different geometric regions by adaptively clustering the candidate corner point number of the image region where the cluster is located, thereby improving the accuracy of shaft part forging quality monitoring.

The application discloses a kind of corner point detection method and its device, the detection method includes the following steps: obtaining the picture to be detected, initializes angle set;The edge of the picture to be detected is detected and the edge graph is obtained, and curve is extracted from the edge graph;Each curve is processed as follows: fill the gap in curve, find several T joints in curve and add all the T joints found to angle set;The curve c j It is handled, the corner point in curve is identified, and the corner point identified is added to angle set, and the corner point in angle set S is the corner point in the picture to be detected.The detection method has the advantages of consistent with real curvature, angle convex / concave small and low computational complexity.

The invention discloses a checkerboard corner detection method, device and equipment and a storage medium, and the method comprises the steps: obtaining a checkerboard image, carrying out the global nonlinear gray correction of the checkerboard image based on the global gray distribution information of checkerboards in the checkerboard image, and obtaining a target checkerboard gray image; performing edge detection on the target checkerboard gray level image to obtain a checkerboard edge image; each checkerboard longitudinal line or checkerboard transverse line in the checkerboard edge image is fitted to obtain a plurality of target edge curves, and each target edge curve corresponds to one checkerboard longitudinal line or checkerboard transverse line; and solving intersection points among the multiple target edge curves to obtain angular point information in the checkerboard image. The checkerboard corner detection precision can be effectively improved.

This invention discloses an eye-tracking method and system based on multimodal corner-eye detection. The method includes: Step 1, acquiring user eye image data in real time and performing image enhancement processing; Step 2, multimodal corner-eye detection, sequentially performing initial detection and verification, refined detection, and iterative optimization detection, to obtain candidate corner-eye points through initial detection and verification, accurately locate the corner-eye point position through refined detection, and iteratively optimize the detection results of each frame through iterative optimization detection; Step 3, dynamic slip detection, including: continuously monitoring the positional changes of the corner-eye points of both eyes relative to the corner-eye points of a reference frame, using a single-threshold dual-level judgment mechanism to determine whether device slip has occurred; and updating the corner-eye point data of the reference frame when device slip occurs, and automatically triggering a recalibration process to update the position of the corner-eye points of the reference frame. This invention solves the problem that existing solutions have shortcomings in device slip detection, which leads to a significant decrease in tracking accuracy.

The application provides a Harris corner point detection method and device. The method comprises the following steps: obtaining a target image to be detected by a processor; using a first detection algorithm with a first scale and a second detection algorithm with a second scale larger than the first scale to preliminarily identify texture possible points from the target image; determining a candidate neighborhood by at least performing expansion based on the preliminarily identified texture possible points; using points in the candidate neighborhood as candidate points to calculate the Harris score of each candidate point; and extracting a preset number of candidate points with the highest Harris score from the candidate points as Harris corner points via a non-maximum suppression method. The detection method provided by the application can reduce the calculation amount and the calculation complexity of Harris corner point detection, ensure the real-time performance of the system, and ensure the final use effect of the detected Harris corner points.

The present application relates to the technical field of defect detection, in particular to a suture scalpel packaging quality detection method based on visual nerve, comprising the following steps: collecting suture scalpel packaging images for corner point detection and gradient screening, obtaining alignment parameters under a unified coordinate system, calculating optical flow displacement initial value and residual error, triggering a depth tracking model to correct displacement when exceeding the limit, cutting ROI sub-image to extract blade edge contour, calculating orthogonal projection length to generate scalpel form markers, extracting texture and shape features to input support vector machine for defect recognition, calling random forest regression model to calculate compensation parameters, and outputting quality detection results. In the present application, coordinate mapping is established through corner point detection and gradient screening, displacement is corrected by combining optical flow residual error and depth tracking, ROI cutting extracts blade edge contour, orthogonal projection quantifies form deviation, texture fusion constructs classification basis, support vector machine recognizes defects, random forest regression compensates parameters, and closed-loop feedback optimizes performance.

The application discloses a multifunctional radar wave element extraction method and system based on a dot matrix model, and the method comprises the following steps: for a multifunctional radar (MFR) pulse sequence, a dot matrix (D) based on a pulse repetition interval (PRI) value, a pulse width (PW) value and a radio frequency (RF) value is respectively constructed according to the threshold values of the corresponding PRI value, PW value and RF value of the MFR pulse sequence; PRI ,D RF ,D PW ; the constructed dot matrix (D PRI ,D RF ,D PW is respectively subjected to image preprocessing, image hole filling and image closing operation; an L-shaped corner point on a main diagonal line of the preprocessed dot matrix is detected by using a Harris corner point detection operator, that is, a wave element starting point; and a union set of the detection results is obtained, that is, a wave element starting pulse set. The application does not depend on cluster parameter setting, can fully utilize the overall modulation law of the wave element, reduces the false alarm probability while improving the wave element detection probability, and still has good performance in the case that the pulse is lost or the parameter measurement error is large.

The present application relates to the field of remote sensing image road extraction, and aims to provide a remote sensing image road extraction missing completion method based on road width reasoning, so as to solve the problem that the road extraction result is incomplete due to the influence of the coverings on both sides of the road in the road extraction based on remote sensing images, and achieve the goal of automatic and accurate road extraction. In order to solve the above technical problem, the specific steps include the following: first, obtaining a remote sensing image to be extracted, and performing road extraction based on a U-net network; then, determining the position of the road intersection based on the road network data of the remote sensing image; second, performing corner point detection on the road extraction result based on a corner point detection algorithm, and matching the detected corner points with the road intersection; finally, performing road width reasoning based on the matching result of the road intersection and the corner points, and completing the road extraction result. The present application fundamentally eliminates the influence of the coverings on both sides of the road on the road extraction, and achieves accurate road extraction.

The invention provides a corner detection method and device. The method comprises the following steps: acquiring a to-be-detected image; inputting the to-be-detected image into a pre-trained corner detection model to obtain a corner detection result; wherein the corner detection model is obtained by training by using a pre-constructed total loss function according to a pre-constructed sample data set based on a deep convolutional neural network; the sample data set comprises a plurality of checkerboard image samples, and the total loss function comprises continuous heat map matching loss. According to the method, the corner detection model is constructed through the deep convolutional neural network, and the corner detection model is trained by using the continuous heat map matching loss in the training process, so that the positioning capability of the model on the sub-pixel precision is remarkably improved. In the reasoning process, an image to be detected is input into the trained corner detection model, a corner detection result is directly obtained, and end-to-end sub-pixel-level fine positioning is achieved.