156 results about "Corner point detection" patented technology

A fully automated package identification and measuring system, in which an omni-directional holographic scanning tunnel is used to read bar codes on packages entering the tunnel, while a package dimensioning subsystem is used to capture information about the package prior to entry into the tunnel. Mathematical models are created on a real-time basis for the geometry of the package and the position of the laser scanning beam used to read the bar code symbol thereon. The mathematical models are analyzed to determine if collected and queued package identification data is spatially and / or temporally correlated with package measurement data using vector-based ray-tracing methods, homogeneous transformations, and object-oriented decision logic so as to enable simultaneous tracking of multiple packages being transported through the scanning tunnel.

A fully automated package identification and measuring system, in which an omni-directional holographic scanning tunnel is used to read bar codes on packages entering the tunnel, while a package dimensioning subsystem is used to capture information about the package prior to entry into the tunnel. Mathematical models are created on a real-time basis for the geometry of the package and the position of the laser scanning beam used to read the bar code symbol thereon. The mathematical models are analyzed to determine if collected and queued package identification data is spatially and / or temporally correlated with package measurement data using vector-based ray-tracing methods, homogeneous transformations, and object-oriented decision logic so as to enable simultaneous tracking of multiple packages being transported through the scanning tunnel.

A fully automated package identification and measuring system, in which an omni-directional holographic scanning tunnel is used to read bar codes on packages entering the tunnel, while a package dimensioning subsystem is used to capture information about the package prior to entry into the tunnel. Mathematical models are created on a real-time basis for the geometry of the package and the position of the laser scanning beam used to read the bar code symbol thereon. The mathematical models are analyzed to determine if collected and queued package identification data is spatially and / or temporally correlated with package measurement data using vector-based ray-tracing methods, homogeneous transformations, and object-oriented decision logic so as to enable simultaneous tracking of multiple packages being transported through the scanning tunnel.

The invention relates to a computer vision based automatic labelling positioning method. The method comprises the following steps: acquiring an image of a labelling sample product through an image acquiring module; preprocessing the product image through an image recognizing module; determining an interesting area based on a position to be labelled; performing corner-point detection in the interesting area in order to recognize pixel coordinates of corner points of a label; finally and accurately positioning pixel coordinates of a center point of the label; calibrating a camera; transforming the pixel coordinates of the center point of a paper carton image into world coordinates through a data processing module in a computer processing controller based on the obtained camera parameters; returning the obtained spatial position of the center point of the label as known information to a manipulator; determining a moving route based on the input position information through the manipulator; and finding out a position point to be labelled so as to realize automatic labelling operation. The method can be continuously carried out for a long time at high speed, high accuracy and high resolution ratio; the recognizing rate is high; and the measurement result is accurate.

The invention discloses a detection method and a detection device for a character region. The detection method for the character region comprises the following steps: carrying out corner point detection on a target image so as to obtain corner point high confidence coefficient regions of the image; carrying out edge detection on the target image so as to obtain an edge communication region of the image; judging whether any detected corner point high confidence coefficient region meets preset regular shape conditions; if yes, determining the region to serve as the character region; if not, further judging whether the coincidence proportion of the region to the edge communication region reaches a preset threshold value; if yes, determining the region to serve as the character region. With the technical scheme, by adopting a manner that the corner point detection is combined with the edge detection, the character region on the image is determined; under the conditions of complicated character background and irregular character regions, the character region can be detected more effectively.

The invention provides a rapid image splicing method based on feature matching. The method includes the steps: respectively extracting line features and point features of images by the aid of a Canny edge detection algorithm and a Harris corner point detection algorithm and combining the line features and the point features to obtain the best feature points; roughly matching the feature points by the aid of similarity metric NCC (normalized cross correlation), removing mismatched points by the aid of an RANSAC (random sample consensus) algorithm to improve image matching accuracy, and calculating transformation model parameters by an LSM (least square method); finally, fusing the spliced images by a weighted average method and eliminating splicing gaps. The points to the matched are determined according to the point and line image features, image details can be enhanced, image matching errors caused by underexposure, overexposure, camera shake and the like are avoided, and image splicing quality is improved to a certain degree.

The invention discloses a method for detecting the distance measured ahead, which includes the following steps: S1. Using a visual sensor to collect image information in the direction of travel of the vehicle, and using machine vision to detect the above-mentioned collected image information to obtain the target in the image information area; S2, in the target area in the image information acquired by machine vision, use the edge feature to extract the specific position of the front vehicle in the image information, and then use the corner point detection method to determine the position of the pixel point of the front vehicle in the image information; S3 1. Calculate the actual distance between the front vehicle and the self-vehicle by using the monocular vision ranging geometric model; the invention can be used to detect the distance of the front vehicle under different driving speeds and different lighting conditions on structured roads, and the invention is universal High, good robustness, fast calculation speed, simple implementation.

The invention discloses a tracking and matching parallel computing method for a wearable device so as to achieve augmented reality tracking and matching. According to the tracking and matching parallel computing method, the SCAAT-EKF feature tracking technology is adopted, complementary fusion data acquisition is conducted on multiple sensors in the wearable device, and data collision can be effectively avoided; the operation strategy based on the double kernal CPU+GPU group kernel multi-channel is utilized, corner detection and extraction based on the Harris algorithm are conducted in the GPU, the double kernal CPU is used for conducting P-KLT tracking and matching calculation, and therefore algorithm fast parallel processing is achieved. The tracking and matching parallel computing method mainly comprises the steps of hybrid tracking and feature extraction for the wearable device, accurate extraction of feature points of target natural features without marks, Harris corner point detection achieved based on a GPU parallel processing mechanism, the CPU-based P-KLT parallel feature tracking algorithm and the secondary matching optimization algorithm. The tracking and matching parallel computing method for the wearable device achieves combination of the sensors of the wearable device and visual tracking and matching, and has the wide prospect in the augmented reality three-dimensional registration aspect.

The invention discloses an image registration method based on improved mutual information and Harris corner point features. The method mainly solves the problems that through a traditional method, registration consumes much time and registration accuracy is low. The method includes the implementation steps that firstly, an image F to be registered and a reference image R are input; secondly, rectangular frames are placed in any position of the image F to be registered and any position of the reference image R, and most similar rectangular frames are obtained; thirdly, corner point features of the most similar rectangular frames are extracted, and matching points in the similar rectangular frames are obtained; fourthly, the similar rectangular frames on which the matching points are extracted are registered after mismatching points are deleted, and registered transformation parameters are recorded; fifthly, affine transformation is conducted on the image F to be registered through the transformation parameters; sixthly, the image obtained after affine transformation is fused with the reference image R, and then a registered image is obtained. The image registration method has the advantages of being short in consumed registration time and high in registration accuracy, and can be used for registration of medical images, natural images and synthetic aperture radar images.

The present invention relates to a parallel dual-camera stereo calibration method. The method comprises the following steps of: (1) providing a checkerboard calibration plate, and taking black squares in four corners of the checkerboard calibration plate as coverage areas; (2) setting the checkerboard calibration plate to have different angles, and shooting the checkerboard calibration plate to obtain multiple sets of left images and right images; (3) processing four coverage areas in each image to obtain four quadrangle areas; (4) positioning the quadrangle areas corresponding to the four coverage areas to acquire coordinates of four vertices in a corner point detection area, so as to obtain the corner point detection area; and (5) obtaining a shift mode during 3D image synthesis by the left and right images shot by a shooting apparatus according to the coordinates of the four vertices in the corner point detection area of left sample images and right sample images, and carrying out calibration on internal parameters and external parameters of a left camera and a right camera.

The invention discloses a method for positioning a DPM two-dimension code area. The method includes the steps that a collected image containing DPM two-dimension codes is subjected to graying processing to obtain a gray image; corner point detection is performed on the gray image through an improved SUSAN corner point detection algorithm, if corner points are detected, a next step is executed, and if no corner point is detected, the procedure is finished; an AP clustering algorithm with a semi supervision and guidance mechanism is added to cluster the detected corner points, and a priori conditions are built to determine clustered results of suspected corner point accumulation areas of the DPM two-dimension codes; clustering areas meeting the a priori conditions are marked as positioning results, a positioning image is output, and the procedure is finished. The method can reduce noise influences to a certain degree, and the phenomenon of pixel missing or pixel redundancecan not happen on a bar code area, so that the foundation is laid for accurately recognizing bar code content. The method achieves the purpose of rapidly and accurately positioning the DPM two-dimension code area and reduces production cost of two-dimensional code recognition products.

The invention discloses an image correction method suitable for a bending deformation two-dimensional code, and specially aims at the problem of two-dimensional code nonlinear deformation because of the photographing angle and the bending surface. The method is mainly divided into two steps of a two-dimensional code area extraction and bending deformation correction. The step of two-dimensional code area extraction comprises two-dimensional code image preprocessing, two-dimensional code position detection pattern positioning and two-dimensional code closed extraction. The step of bending deformation correction comprises original image reference corresponding point extraction, standard image reference corresponding point extraction and construction of a bending deformation correction transformation model. The two-dimensional code corner point detection method is improved by using the slope and intercept knowledge of plane geometry to realize accurate positioning of the two-dimensional code corner points, a "roller" method is designed to solve the edge points of the bending deformation two-dimensional code, and the side length relationship of the two-dimensional code and the positiondetection pattern is calculated through combination of curve fitting to construct the transformation model so that adaptive correction of the bending deformation two-dimensional code under the general condition can be realized.

The invention discloses a method for identifying honeycomb edges from honeycomb core surface measurement data. The method comprises the following steps: performing two-dimensional coordinate transformation on collected honeycomb core surface three-dimensional data, identifying corner points in a two-dimensional plane projection image of a honeycomb core through a corner point detection algorithm,wherein the corner points comprise a human-shaped vertex, a Y-shaped vertex, a pseudo vertex and unrecognized missing corner points; based on the proposed corner type judgment algorithm, sequentiallyidentifying two endpoints of the honeycomb edge, determining one endpoint during adjacent edge identification, and determining the other endpoint through local analysis of the identification vertex; and effectively eliminating pseudo vertexes in the implementation process of the honeycomb edge, supplementing missing vertexes, and achieving stable and high-precision honeycomb edge recognition. According to the method, the data processing problem of honeycomb core measurement is effectively solved, and the method is flexible, high in precision and good in robustness, and the measurement requirement of the honeycomb core surface after actual machining can be met.

The invention relates to a tooth error vision measurement method for the involute of a straight toothed spur gear. According to the method, an involute tooth error measurement model is established based on a linear laser vision imaging system. In combination with the light stripe centric line detection technique, the corner point detection technique, the camera calibration technique and the like, the measurement of a to-be-measured object is realized. The method comprises the steps of 1) calibrating a vision imaging model under the effect of laser active illumination; 2) calibrating the parameters of a pseudo light plane equation; 3) calculating the three-dimensional coordinates of a to-be-measured point on the surface of a tooth profile; 4) calculating the tooth error for the involute of the surface of the tooth profile. According to the method, based on the vision measurement technique, the online non-contact detection is implemented. Meanwhile, based on the precise laser active illumination technique, the high measurement accuracy is ensured. The measurement accuracy of the method for gears of IT 6-8 level in accuracy is smaller than or equal to 0.015 mm.

The embodiment of the invention discloses a plane calibration board and a calibration data acquisition method and system, and the plane calibration board comprises a plurality of separate checkerboardunits which are distributed in an array manner. A preset distance is reserved between every two checkerboard units; and at least three characteristic hollow circles which are respectively arranged inthe three different checkerboard units in the preset range of the central area of the calibration board. According to the scheme, the four edges or the four vertexes of the checkerboard unit are usedfor restraining the corner points, and the stability and precision of corner point detection are improved. And meanwhile, chessboard positioning information is added by utilizing the characteristic hollow circle, so that the plane calibration board has a directional characteristic. In addition, two-dimensional code information is added to part of the checkerboard units, so that the initial positions of the checkerboard units in the camera can be positioned. Even if a part of checkerboard units are shielded or exceed the field of view of the camera, the camera can be normally calibrated, calibration and image acquisition are more convenient and flexible, and particularly, corner sampling of a large-distortion camera is facilitated.

The invention relates to a fisheye-camera calibration system based on the genetic algorithm and an image distortion correction method thereof, and belongs to the technical field of computer graphics, digital image processing and the artificial intelligence algorithm. The fisheye-camera calibration system comprises a distortion correction module, a feature detection module and a fisheye-camera parameter calibration module which are sequentially connected. The image distortion correction method includes the steps of image pretreatment, Harris corner point detection and genetic-algorithm internal-reference and external-reference approaching. By means of the fisheye-camera calibration system based on the genetic algorithm and the image distortion correction method thereof, calibration parameters can be reduced, and the accuracy of the calibration internal reference and the accuracy of the calibration external reference are increased.

The invention discloses a fisheye camera calibration method and device. The method comprises steps that a world coordinate system is constructed in the space where a calibration plate is located, andthree-dimensional coordinate values of inner corner points on the calibration plate are further determined; an image of the calibration plate taken by a fisheye camera is acquired; corner point detection of the image is carried out through the calibration algorithm, and corresponding three-dimensional space coordinate values are determined according to image coordinate values of all the detected inner corner points or a part of the inner corner points and are further stored; internal and external parameters of the fisheye camera are calculated based on point pairs of the image coordinate values and the three-dimensional coordinate values including the inner corner points, and camera calibration is completed. The fisheye camera calibration device comprises a coordinate system construction unit, an image acquisition unit, a detection corresponding unit and a calibration unit. The method is advantaged in that flexible calibration is carried out according to whether all the inner corner points of the calibration plate can be detected, calibration efficiency is improved, and large-scale industrial implementation of fisheye camera calibration is facilitated.

The embodiment of the invention discloses a parking space detection method and device, equipment, a storage medium and a vehicle, and relates to the technical field of autonomous parking. The parkingspace detection method comprises the steps of obtaining an input image for presenting a to-be-detected target parking space; and simultaneously detecting a corner point of the target parking space anda parking space line using the input image, and then correcting a position of the detected corner point based on the detected parking space line. According to the embodiment of the invention, the corner point and the parking space line of the target parking space in the image are detected at the same time, and the corner point is corrected by using the relatively stable parking space line, and the joint detection method can improve the corner point detection precision and reduce the shaking of the corner point. Besides, some embodiments of the invention can use the neural network model to complete parking stall detection so as to improve the detection speed and save the detection time consumption.

The invention relates to a fold distortion QR code correcting method. The method comprises three parts of two-dimensional code corner point detection, linear deformation correction and fold correction. If one pair of sides of a two-dimensional code pattern are not distorted and the other pair of sides are distorted, linear deformation effect caused by shooting angle and the like is eliminated to acquire a first corrected image, and then a shortest curve external connection method is used to carry out fold distortion correction on the first corrected image to acquire a second corrected image. On the basis of the second corrected image, a QR code is decoded and recognized. The step of fold distortion correction comprises edge contour extraction, edge contour interference electricity removing, edge contour multi-function fitting, fitting function screening, internal point mapping function selection, mapping transformation correction and the like. According to the invention, the detectionspeed and the recognition rate are greatly improved, and the fold distortion two-dimensional code is corrected and recognized.

The invention relates to a curve corner point detection method based on difference accumulated values and three-point chain code differences, which is technically characterized by comprising the following steps of: firstly carrying out Freeman chain code encoding to the obtained curve, calculating the difference accumulated values of the chain codes point by point according to the Freeman chain code encoding sequence based on the obtained Freeman chain codes, and dividing all pixel points on the curve into determining points, non-corner points and doubtful points through the difference accumulated values of the three-point chain codes; then calculating the three-point chain code differences of the doubtful points, and judging the doubtful points through the three-point chain code differences to find out true corner points in the doubtful points; and finally merging the corner points found in the determining points and the doubtful points to obtain all corner points of the curve. In the method, the points with curvatures to be calculated are judged by adopting the three-point chain code differences to replace the curvatures, thereby further reducing operation workload and lowering operation complexity. The corner points detected in the method also contain sequence characteristics which provide convenient for utilizing the corner points to carry out image processing.

The invention provides an anti-counterfeit identification method based on hybrid stacking of a plurality of characters and numbers of a two-dimensional code. The method comprises the steps of calculating a communicating threshold, extracting a position detecting pattern area according to gravity center detection and area ratio, obtaining four corner points of the two-dimensional code through a line segment with longest distance among three position detecting patterns, eliminating influence of two-dimensional deformation and photographing angle to corner point detection of the two-dimensional code, respectively calculating an innermost square area, a middle square area and an outer square area of the position detecting pattern according to the center of the position detecting pattern by means of three times of area increase, calculating the corner points of the outer square area of the position detecting pattern by means of approximation of two sets of lines, calculating the edge key points of four boundaries of the two-dimensional code by means of a bridging method, calculating the length of the curve between the key points of the deformed two-dimensional code by means of integration, and calculating the corresponding coordinates of the key points on the deformed two-dimensional code on the corrected image through a telescoping ratio. The anti-counterfeit identification methodis a new anti-counterfeit identification method and effectively protects related knowledge property and economic benefit of legal manufacturers.

The invention belongs to the technical field of image processing, and particularly discloses a checkerboard corner detection method based on a neural network and application of the checkerboard cornerdetection method. The method comprises the following steps: acquiring a multi-angle image of a checkerboard to be detected, and preprocessing the multi-angle image to obtain a grayscale image; basedon a neural network, carrying out angular point detection on the gray level image so as to determine candidate angular points; removing the candidate angular points which do not meet the geometric standard of the checkerboard to determine the angular points to be detected; determining sub-pixel positions of the checkerboard corner points to be detected according to the obtained corner points to bedetected; and performing region screening on the obtained to-be-detected corner points, determining an index matrix of each to-be-detected corner point, and further obtaining a plurality of checkerboard corner point sequences which are not crossed with one another, thereby finishing corner point detection of the to-be-detected checkerboard. The method has high detection robustness for interference such as illumination shielding and noise offset, has high positioning stability for image bending and distortion caused by uneven lenses or wide-angle camera shooting, and can complete detection formultiple chessboards at the same time.

The invention discloses a rectangular pin element coarse positioning method based on corner points, which belongs to the technical field of image processing, and relates to an element positioning method, and solves the problems that the element positioning algorithm in the existing surface mounting machine is poor in universality, low in execution efficiency and sensitive to lighting. Element template information is obtained by using the image processing method or an element data sheet, a template coordinate system is built, a distance matrix and an angle matrix for key vertexes of the element template pin, and the quadrant of the key vertexes in the template coordinate system are extracted; corner point detection is carried out on a to-be-positioned element picture, and coarse correspondence between the key vertexes and the corner points of the element template pin is carried out; each key vertex and corresponding multiple corner points are acquired; aggregation is carried out on each key vertex and the corresponding multiple corner points, the corner point positions obtained through aggregation are used for extracting subpixel corner points, the center and the angle of a chip are obtained according to the position of each subpixel corner point, and rectangular pin element coarse positioning based on the corner points is completed. The method of the invention is suitable for being used for element positioning.

The invention relates to a method for detecting unmanned aerial vehicle visual navigation landing cooperative target robust, which is used for solving the problem that a cooperative target is difficult to detect accurately due to the large distortion of a cooperative target which is shot by an airborne camera and a plurality of non-cooperative targets in an image. A red square is designed to be a large back plate, fluorescent green H and small triangles are used as the cooperative targets, a corner point detection algorithm which combines a massively parallel processor (MPP) method and a Harris method is proposed to accurately detect the corner points of the cooperative targets under the condition of complicated environments and large distortion, and the method effectively combines the advantages of high detection precision of a Harris corner point and directionality of an MPP corner point. Compared with the conventional scheme for detecting an unmanned aerial vehicle visual navigation landing cooperative target, the method is not only easy to realize, but also can be used for effectively solving the problem of detecting the cooperative targets under the condition of complicated backgrounds and large distortion. The detection accuracy is improved, the engineering is conveniently realized, and the method has an important meaning to unmanned aerial vehicle independent safety landing.

The invention discloses a system and method for reconstructing a 360-degree panoramic video image, and the method comprises the following steps: solving the internal and external parameters and distortion coefficients of four fisheye cameras through employing a Zhang Zhengyou calibration method based on the detection of black and white checkerboard internal corners of a calibration board; then, based on external corner point detection of the square mark block graph, solving to obtain external parameters of the four fisheye cameras under a world coordinate system; and finally, according to theinternal and external parameters and the distortion coefficient of the fisheye camera, utilizing the characteristic that the relative position relationship among the four distributed fisheye cameras is fixed, enabling a four-core processor based on an NXP S32V234FS to adopt a space-time overlapping parallel pipeline processing mechanism to perform splicing and fusion processing on video images shot by each fisheye camera, and forming a 360-degree panoramic video image in a world coordinate space; the method has the characteristics of high cost performance, high reliability, high convenience and the like.

The embodiment of the invention discloses a checkerboard corner point detection method and device. The method includes the steps that a gray image of a calibrated scene is obtained, and candidate corner points are determined according to the gray image; corner point growth is conducted on the candidate corner points through the corner point growth algorithm, the candidate corner points are screened according to corner point growth results, and checkerboard corner points are determined; sub-pixel positions of the checkerboard corner points are determined according to the checkerboard corner points. The method and device solve the problem that symmetric background corner points can not be eliminated, precise corner point detection can be conducted under a complex calibration background, and the robustness of corner point detection is improved.

The invention belongs to the field of image identification, and provides a checkerboard pattern corner point detecting method and a checkerboard pattern corner point detecting device. The method comprises the following steps that: checkerboard pattern images are obtained; the obtained checkerboard pattern images are subjected to corner point enhancement filtering to obtain enhanced checkerboard pattern images; located target regions of corner points are determined in the enhanced checkerboard pattern images; and coordinates of the corner points in each target region are determined. The method and the device provided by the invention have the advantages that a method based on polar coordinate shape descriptors and automatic mode identification is adopted for checkerboard pattern corner point detection, and the problems of high requirement on image quality, poor antinoise performance and low corner point detection precision of a conventional corner point detecting method are effectively solved.

The invention discloses a machine visual identification method for sorting products with corner point characteristics. The method comprises the steps as follows: performing Freeman chain code encoding on the image outlines of known products, detecting the corner points by using the Freeman chain codes, calculating the corner point characteristic three-dimensional column vectors of all the corner points, sequencing the corner point characteristic three-dimensional column vectors and taking the vector sequence as the vector sequence of the known products; performing Freeman chain code encoding on the image outlines of objects to be sorted, detecting the corner points by using the Freeman chain codes, calculating the corner point characteristic three-dimensional column vectors of all the corner points, sequencing the corner point characteristic three-dimensional column vectors and taking the vector sequence as the vector sequence of the objects to be sorted; and judging the vector sequence of the known products and the vector sequence of the objects to be sorted and finishing machine visual identification. According to the method, the shapes of the known products and the objects to be sorted are simply represented by the chain codes; the subsequent calculation quantity is small; the extracted vector sequence has zooming and rotation invariance can be accurately identified; and the algorithm is simple.

The invention provides a people counting method, a people counting system and electronic equipment. According to the people counting method, the people counting system and the electronic equipment provided by the embodiment of the invention, corner point detection is performed on an image to be processed, and then the whole image is divided into multiple local sub-images according to density distribution of detected corner points, so that people counting is performed based on each local sub-image, and the accuracy of people counting can be improved.

The invention relates to intelligent traffic systems, and provides a detection method for a front automobile window top right corner point. The detection method for the front automobile window top right corner point detection method comprises the steps of selecting a sample set and manually calibrating a number plate center point and the front automobile window top right corner point, looking for the relation between the number plate center point and the front automobile window top right corner point, setting a sample base, training an adaboost classifier, locating the number plate center point, locating the front automobile window top right corner point, determining a searching area, collecting a candidate area, extracting features, and introducing the trained adaboost in for classification. The front automobile window top right corner is located, and therefore position reference is provided for other detected portions; the algorithm extracts the automobile window features and puts the automobile window features into the adaboost classifier to be trained, the detection rate of an automobile window can be improved, the influence to the result by quality of the selected sample set and the features is large, and the detection method for the front automobile window top right corner point can achieve the good accuracy when good samples are selected.