24 results about "Edge tracking" patented technology

The invention discloses a coal mine belt deviation real-time lightweight detection device and a coal mine belt deviation real-time lightweight detection method based on image sub-pixels, firstly, the device captures video streams of belt operation in real time through an image acquisition device, analyzes the video streams in real time by means of edge computing nodes, and monitors the running state of a belt in real time to determine the deviation of the belt; once the belt starts to work, image distortion caused by a shooting angle or environment is immediately corrected through perspective transformation. Secondly, automatically selecting a high-contrast reference strip through an image preprocessing method, and accurately extracting a region near the belt edge as a minimum region of interest, namely a belt edge region; then, a lightweight sub-pixel convolutional neural network is used to carry out deep analysis on the areas, and a high-quality super-resolution image is generated, so that the accuracy of edge detection is further improved. And finally, accurately identifying and positioning the edge position of the belt by introducing a bidirectional edge tracking method with an abnormal detection function, thereby realizing real-time and accurate detection on the deviation condition of the belt.

The invention relates to an image edge extraction method based on Canny detection operator optimization, and belongs to the technical field of image detection. The method comprises the following steps: acquiring a source image, converting the source image into a grey-scale image, performing down-sampling operation on the grey-scale image to obtain a down-sampling image, filtering the down-sampling image, and calculating a gradient magnitude and a direction of the filtered image to obtain a gradient magnitude image; removing non-edge points from the gradient magnitude image by optimizing a non-maximum suppression process, and storing candidate image edges; defining a real edge and a weak edge by setting a careful extraction interval, and setting a candidate image edge of which the gradient value is lower than the lower bound of the careful extraction interval as 0; lagging edge tracking is carried out on the weak edge to judge whether the weak edge and the strong edge are continuous or not, if yes, the weak edge is set as a real edge, and if not, the weak edge is removed. The accuracy and robustness of edge detection are remarkably improved in an image with more noise, and the phenomena of generation of false edges and edge breakage are reduced.

The application discloses a kind of liquid bag film loading conveying devices, belong to infusion bag production equipment technical field.The film coating device of the present application includes rack and the main material roller sequentially arranged in rack;Tension control mechanism, including lifting piece and multiple with film conveying direction equidistant arrangement roll group, each described roll group includes up and down misregistration fixed roller and adjusting roller, the fixed roller with the rack is fixed, the lifting piece can drive the adjusting roller with the direction perpendicular to film conveying direction displacement, and displacement amplitude and the rotation torque of the main material roller are matched by electric control adjustment;Mould closing mechanism, to the multiple described film is flattened after mould closing;Edge tracking mechanism, including bearing table, double-shaft displacement piece, first pressing die set, second pressing die set, the film after mould closing is supported by the bearing table, and edge tracking is carried out by first pressing die set and the second pressing die set, the present application can eliminate film transmission process tension change and film wrinkle.

The invention discloses an industrial CT image wavelet transformation contour extraction method and system based on man-machine interaction. The system comprises a man-machine interaction module, a starting point determination module and a contour extraction module value. The man-machine interaction module specifies an external area point and an internal area point of the area to be extracted. The starting point determination module can determine a contour starting point corresponding to a to-be-extracted region by adopting a wavelet transform module value matrix through specified external region points and internal region points. And finally, the contour extraction module extracts other contour points of the to-be-extracted region according to a set screening rule based on the contour starting point of the to-be-extracted region, and forms a contour corresponding to the to-be-extracted region according to all the extracted contour points. By adopting human-computer interaction, clear prior information can be applied to actual contour extraction, a thresholding screening process is removed, edge positioning of the contour is improved, additional edge tracking is not needed any more, edge extraction and edge tracking are combined into one, and operation steps are simplified.

1. The name of the design product: rectifier (automatic edge tracking). 2. The use of the design product: automatic edge tracking of material rectification sensor for wide equipment, rectification sensor and material follow-up. 3. The design points of the design product: in shape. 4. The picture or photo that best indicates the design points: assembly 1 perspective view. 5. Other circumstances that need to be explained: assembly 1 and assembly 2 are complete design, used together.

The invention relates to the technical field of display screen detection, in particular to a multi-spectral imaging-based display screen pixel-level defect dynamic detection method, which comprises the following steps of: acquiring high-resolution images of a plurality of fixed wavebands during dynamic display of a display screen by adopting a multi-waveband parallel imaging mode; filtering edge tracking and fuzzy region interpolation compensation are carried out, dynamic threshold updating is adopted, and a single physical pixel is segmented in a dynamic scene; extracting three types of multi-dimensional features of multi-band light intensity, band association and space-time dynamic of pixels, identifying and classifying pixel-level defects of the display screen by adopting multi-band cross validation, and outputting identification and classification data; through the above mode, the dynamic adaptation and segmentation precision is improved, and the large-scale production quality management and control requirements of high-resolution and high-refresh-rate display screens are met.

This invention relates to the field of display screen inspection technology, specifically to a dynamic detection method for pixel-level defects in display screens based on multispectral imaging. The method includes the following steps: employing multi-band parallel imaging to acquire high-resolution images of multiple fixed bands during dynamic display; performing filtered edge tracking and blurred region interpolation compensation, and using dynamic threshold updates to segment individual physical pixels in dynamic scenes; extracting three multi-dimensional features of pixels—multi-band light intensity, band correlation, and spatiotemporal dynamics—and using multi-band cross-validation to identify and classify pixel-level defects in the display screen, outputting the identification and classification data; through the above methods, dynamic adaptation and segmentation accuracy are improved, meeting the quality control requirements of large-scale production of high-resolution, high-refresh-rate display screens.

The present invention discloses a method for target segmentation completion and recognition in an occluded environment, including: collecting two consecutive video sequence images in an occluded environment, converting the images from the RGB color space to the YCbCr color space, extracting the Y component and performing normalization processing; performing mean clustering segmentation to obtain two segmentation images in the occluded environment; performing edge extraction and edge tracking, implementing contour reconstruction of the picking target through an improved interpolation algorithm, and then calculating and generating a fitting rectangle according to the segmentation images with contour reconstruction in the two occluded environments; calculating the relative accuracy, identifying the occluded repair image, and obtaining the segmentation completion and recognition image of the picking target in the occluded environment. The present invention provides an effective method for target segmentation completion and recognition in an occluded environment, and innovatively uses an improved interpolation algorithm to complete the contour of the picking target, having the advantages of strong practicability, high contour completion accuracy, and strong anti-background environment interference ability.

This invention provides a multi-target tracking method and system for special steel production processes using edge-cloud collaboration, belonging to the field of industrial intelligent manufacturing. The method deploys a multi-target tracking model at the edge of the special steel production line and a multimodal inference model in the cloud. Real-time production data is collected and input into the multi-target tracking model, outputting initial tracking results. Based on the initial tracking results, a scenario complexity index is calculated, and then it is determined whether the scenario meets stability conditions. If it does, the initial tracking result is output as the edge tracking result; otherwise, the cloud-based multimodal inference model performs multimodal inference based on the data uploaded from the edge side to obtain the cloud tracking result. The edge side fuses the edge and cloud tracking results to obtain and maintain the multi-target tracking trajectory. The cloud uses knowledge distillation to train a copy of the multi-target tracking model and distributes it to the edge side to update the model. This invention improves the accuracy and precision of multi-target tracking in special steel production processes.

The present application provides a structural displacement measurement method and system based on video interpolation and computer vision. The structural displacement measurement method acquires structural vibration video data through video interpolation technology based on computer vision, and introduces an attention mechanism into the video interpolation algorithm to obtain an improved video interpolation algorithm. By shooting the vibration video of a cantilever beam model as a data set, the performance of the improved video interpolation algorithm in structural vibration testing is evaluated. Finally, based on the interpolated structural vibration video, the time history data of the measured point displacement is obtained by detecting in the measurement area through edge detection technology and edge tracking technology based on the local area effect, the vibration response of the laboratory model is monitored, and modal parameter identification is carried out. The structural displacement measurement method processes the inter-frame relationship and weight allocation more comprehensively and accurately through the improved video interpolation algorithm, improving the quality of the generated intermediate frames and the monitoring accuracy of structural vibration displacement.

A high-resolution remote sensing image edge repairing method based on tracking, first, a deep convolutional neural network is designed according to the remote sensing edge detection task, samples are made and edge detection model and edge tracking model are trained; the remote sensing image is input into the edge detection model to extract the ground feature edge, so as to obtain the ground feature edge intensity map. The remote sensing image and the random starting point file are input into the edge tracking model to iteratively generate the edge tracking graph, and for the discontinuous ground feature edge, the edge detection result and the edge tracking result are fused to automatically determine the edge width and complete the edge. The method completes the broken line repairing work through the topological information provided by the edge tracking algorithm for the discontinuous edge part in the prediction result of the edge detection algorithm, and simultaneously uses the edge binary graph to assist in connecting the broken edge tracking graph, or creates an edge in the position where the edge does not exist to complete the edge tracking graph. The method can improve the edge extraction accuracy and efficiency, and improve the accuracy of the edge tracking model.

This invention relates to an edge detection and tracking method. The edge detection stage includes color grayscale conversion, Gaussian filtering, generation of gradient magnitude matrix, generation of gradient magnitude histogram and calculation of threshold. The edge tracking stage includes initial point selection, edge concatenation, removal of redundant points, and edge verification. The input to this invention is a color photograph image, and the output is a set of contour line information contained in the image. It is particularly suitable for real-time algorithms in embedded hardware development and can be applied to practical scenarios such as educational robots, pattern recognition, and autonomous driving.

The application discloses a card ticket image segmentation method, device, equipment and program product, and is used for solving the problems of inaccurate edge positioning, poor adaptability and error accumulation in a complex scene. The scheme comprises the following steps: determining an edge starting point of a target card ticket image. Starting from the edge starting point, performing multiple rounds of edge tracking on the target card ticket image to determine an edge path; wherein each round of edge tracking comprises the following steps: taking a current edge point as a center, intercepting a local image in the target card ticket image, and generating a current state vector based on the local image; inputting the current state vector into a first intelligent agent to predict multiple candidate positions of a next edge point; inputting the current state vector into a second intelligent agent to predict an offset direction from the current edge point to the next edge point; and determining the next edge point based on the offset direction and the multiple candidate positions determined in the current round. A segmentation mask is constructed based on the determined edge path, and the target card ticket image is segmented based on the segmentation mask.

Techniques and systems are provided for edge detection is provided. The process can include obtaining a depth information and location information for a point detected in an obtained image; predicting a pose for the point based on the depth information, location information, and information about movement of a device; determining a search range and a search direction based on the location information and predicted pose; performing normalized cross correlation (NCC) matching between pixels of the obtained image based on the determined search range and search direction to identify an edge; and outputting a location of the edge.

Disclosed are a semi-automatic image data labeling method, an electronic apparatus and a non-transitory computer-readable storage medium. The semi-automatic image data labeling method may include: displaying a to-be-labeled image, the to-be-labeled image comprising a selected area and an unselected area; acquiring a coordinate point of the unselected area and a first range value; executing a grabcut algorithm based on the coordinate point of the unselected area and the first range value acquired, and obtaining a binarized image divided by the grabcut algorithm; executing an edge tracking algorithm on the binarized image to acquire current edge coordinates; updating a local coordinate set based on the current edge coordinates acquired; updating the selected area of the to-be-labeled image based on the local coordinate set acquired.

The present application aims at the problem that the existing image edge detection relies on pixel-level edge and has low detection accuracy, and provides a sub-pixel edge tracking method, device and medium based on gray moment model. The method comprises the following steps: obtaining a gray image of a target to be detected and calculating a gradient image; determining edge anchor points based on an initial gradient threshold T1; tracking sub-pixel edges of the image for all edge anchor points; selecting a sub-pixel calculation region, and performing convolution calculation combined with a mask to obtain region image moment values; calculating sub-pixel distances and angles of the edge anchor points to determine sub-pixel positions; selecting a next edge tracking point in the pixel points in the neighborhood centering on the currently selected edge anchor point, and if the gradient of the next edge tracking point is greater than a preset final gradient threshold T2, the next edge tracking point is taken as an edge anchor point for continuous tracking; until the traversal of all edge anchor points is completed, all edge tracking points are obtained; and all edge tracking points are sorted as sub-pixel edge tracking results. The present application is used for sub-pixel level tracking of image edges.

The application relates to the polyester tow twist detection technical field and discloses a polyester tow twist detector, which comprises a fixing frame, a conveying mechanism, a first electric push rod and a limiting mechanism. The conveying mechanism is provided with two groups and is symmetrically arranged in the fixing frame. The polyester tow twist detector is provided with a camera and a limiting mechanism. The camera cooperates with an image processor to realize edge tracking of the tow. Compared with traditional visual interpretation, the detection error of the untwisting angle is reduced, and the detection repeatability is improved. The light supplement lamp ensures stable images within the illumination range, avoids environmental light interference, and integrates a pressure sensor. When the first electric push rod and the second electric push rod push down the two groups of sleeves, the limiting plates and the limiting blocks at the ends of the two groups of sliding rods contact the tow, and the pressure sensor monitors the pressure value in real time. The controller dynamically adjusts the stroke of the first electric push rod and the second electric push rod, so that the clamping force is stabilized in a proper range, and the tow is prevented from being damaged.

Techniques for identifying and eliminating overlaps of characters in an image that are to be processed for character recognition involve selecting a line of contours within the image, generating a line boundary around the line, identifying respective starting edgepoints on each edge of a contour that intersects the line boundary, and tracing a section of the contour by following each edge of the contour until a stopping point indicating an end to the tracing of the section of the contour is reached. The stopping point may comprise a determination that either (i) an end of the section of the contour is reached, or (ii) the section of the contour comprises a first sub-contour, and an intersection between the first sub-contour and a second sub-contour is reached. Based on the determination, the tracing of the section of the contour is completed, and the traced section is deleted from the image.

The application discloses a disinfection door spray precise positioning and identification method based on machine vision, image acquisition of passing crowds is carried out through a high-speed camera to obtain high-quality images; a wavelet threshold denoising method is used to carry out denoising treatment on initial images, and then a weighted average method is used to carry out gray-scale processing; then, the images are subjected to binarization processing; a sobel operator is used to carry out image edge detection and segmentation, and a double-edge tracking algorithm is used to concatenate target edges; then, it is judged whether the images meet spray requirements; if yes, an optimization algorithm is used to calculate the spray time, spray pressure and spray angle of each spray head of N ultrasonic spray nozzles installed on the disinfection door; then, spray disinfection operation is carried out; finally, image processing is used to detect the spray concentration, the nozzle input is adjusted, and the working efficiency and precision of the disinfection door are ensured, so that strong support is provided for inhibiting virus transmission.

The invention relates to the technical field of agricultural production, in particular to an unmanned intertillage weeding agricultural machine navigation method based on visual identification and machine learning, which comprises the following steps: through automatic visual identification and tracking control, acquiring farmland images in real time by using a camera, accurately identifying seedling positions based on a pre-training model and generating a tracking line; autonomous navigation and accurate row control of the agricultural machine are realized; meanwhile, through ridge recognition and edge tracking functions, automatic turning and continuous operation are further supported, the operation precision and efficiency are remarkably improved, manpower dependence and operation difficulty are reduced, and the safety and automation level of agricultural operation are enhanced; in this way, the technical problems that in the prior art, semi-mechanical equipment needs to be manually driven and controlled, the technical requirement for operators is high, the operation line spacing is not accurately controlled, and potential safety hazards are caused due to long-time operation fatigue are solved.

The invention relates to the technical field of image processing, in particular to a grouting anchor pipe welding end welding quality detection method which comprises the steps that a welding gray level image is obtained, a plurality of anchor points in the welding gray level image are obtained, and the texture structure difference of each anchor point is obtained; the method comprises the steps of obtaining a texture normal range interval, screening anchor points according to whether the texture structure difference of each anchor point belongs to the texture normal range interval or not, obtaining a plurality of edge points, carrying out edge tracking on each edge point, if a weld beading exists in the tracking process, obtaining the spanning length according to the protruding condition of the image weld beading, and spanning the weld beading. According to the method, the two continuous edge lines are obtained, whether the welding end and the rod body part are coaxial or not after welding is judged according to the two edge lines, then the welding quality is judged, the complete edge lines are obtained, and the influence of the weld beading edge is avoided.

The present application relates to the technical field of image processing, more particularly, the present application relates to a kind of grouting anchor pipe welding end head welding quality detection method, the method comprises, obtaining welding gray image, obtain several anchor points in welding gray image, obtain the texture structure difference of each anchor point;Texture normal range interval is obtained, according to whether the texture structure difference of each anchor point belongs to texture normal range interval, the anchor point is screened, obtains several edge points, each edge point is edge tracking, if there is welding tumor in the process of tracking, according to the convex situation of image welding tumor, the length of span is obtained, the welding tumor is crossed, obtains two continuous edge lines, according to two edge lines, whether the welding end head and the rod body part are coaxial after welding is judged, and then the welding quality is judged, the complete edge line of the present application is obtained, and the influence of welding tumor edge is avoided.

A system for applying material to a substrate includes a nozzle, robot, actuator, sensor, and controller. The robot provides primary movement such that the nozzle traverses a predefined global bead path across and spaced apart from the substrate. The actuator provides secondary relative movement between the nozzle and substrate. The sensor senses a first location on the substrate that is spaced apart from a location where the nozzle deposits material by a distance based on a sensor response time and relative speed of the nozzle and substrate. The controller detects a feature of the substrate based on the data received from the sensor. The controller directs the actuator to provide the secondary relative movement such that a bead of material is applied to the substrate along a feature-relative bead path. The controller controls the actuator to provide the secondary relative movement based on the response time and the relative speed.

Techniques for identifying and eliminating overlaps of characters in an image that are to be processed for character recognition involve selecting a line of contours within the image, generating a line boundary around the line, identifying respective starting edgepoints on each edge of a contour that intersects the line boundary, and tracing a section of the contour by following each edge of the contour until a stopping point indicating an end to the tracing of the section of the contour is reached. The stopping point may comprise a determination that either (i) an end of the section of the contour is reached, or (ii) the section of the contour comprises a first sub-contour, and an intersection between the first sub-contour and a second sub-contour is reached. Based on the determination, the tracing of the section of the contour is completed, and the traced section is deleted from the image.