30 results about "Blob detection" patented technology

The invention relates to a polylactic acid test piece internal defect detection method, and belongs to the technical field of target detection, the method comprises the following steps: S1, image processing to obtain an input image; s2, attention enhancement feature extraction: preliminarily extracting spatial features, capturing global information and local information, focusing a defect area, and generating an enhancement feature map; s3, carrying out feature fusion, and outputting a fused feature map; s4, performing self-attention re-parameterization, and outputting a local feature map; s5, HSV spot detection including a color conversion algorithm and a defect enhancement algorithm; and S6, outputting a result, establishing a bounding box regression task, predicting a frame coordinate, outputting a defect position, identifying a category label, outputting a defect type, and evaluating reliability and outputting confidence. According to the method, feature extraction is carried out by using depth separable convolution, the visibility and detectability of low-contrast defects are enhanced, and the attention enhancement network is established to adaptively pay attention to key areas related to the defects, so that defect detection is more accurate and efficient.

The invention relates to a method and apparatus for high resolution total body photography (TBP). The inventive method uses a multi-stage keypoints focused pipeline that begins with blob detection to rapidly and coarsely localize potential lesions within high-resolution TBP images. Once these regions of interest are identified, a deep learning classifier evaluates them for malignancy risk. Acknowledging that wide-field imaging can compromise classification precision due to variability in resolution and appearance, the process is further refined by integrating ugly duckling analysis and t-SNE clustering. The ugly duckling detection process groups suspicious regions across all images, effectively highlighting clusters of high-risk candidates for further clinical review utilizing the method of the invention.

This invention relates to the field of autonomous docking and recovery of underwater autonomous vehicles, and in particular to a method for underwater docking guide light identification and positioning for unmanned underwater vehicles. The method includes the following steps: S1, real-time acquisition of guide light image sequences, classification and filtering using an image classification model to obtain pre-filtered guide light images; S2, guide light signal spot detection and brightness-based noise removal to obtain spot detection information; S3, determination of whether the image contains apparent guide light signals to obtain actual guide light spot detection information; S4, calculation of the relative position between the underwater vehicle and the docking station based on the guide light spot information obtained in step S3. A lightweight neural network model is used for real-time optical image recognition, improving the extraction capability of optical feature information, reducing environmental noise interference, and increasing the success rate of guide light identification, positioning accuracy, and docking success rate.

Methods, systems, and media for validating objects within a container, such that a validation system confirms that objects within a container are valid i.e., ready to be shipped from a data center for reuse or ready to be destroyed. The validation system can be used to identify objects, e.g., media, disks, etc., and determine if valid objects are at the correct processing location to be shipped or destroyed. The methods may include identifying the objects, locating an identifier on the object, comparing the identifiers to known identifiers and determining, based on the comparison, if valid objects are at the correct processing location to be shipped or destroyed. The validation system may also use image processing techniques, such as blob detection, to identify invalid objects in the container. The system may reject the entire container if invalid objects are detected.

A system and method for enabling management of faults relating to a tracking managing subsystem and a tracking assembly communicatively connected thereto, include a method including: obtaining from the tracking assembly a tracking stream, which comprises tracking data maps, obtaining a set of system parameters, deriving a set of blobs by applying a blob detection procedure to at least two tracking data maps of the tracking stream, and sending the set of blobs and the set of system parameters to a rendering subsystem at which the set of system parameters are processed, thereby enabling fault management by the rendering subsystem.

A system for blob detection using deep learning is disclosed. The system may include a non-transitory computer-readable storage medium configured to store a plurality of instructions thereon which, when executed by a processor, cause the system to train a U-Net and generate a probability map including a plurality of centroids of a plurality of corresponding blobs, derive two distance maps with bounded probabilities, apply Difference of Gaussian (DoG) with an adaptive scale constrained by the two distance maps with the bounded probabilities, and apply Hessian analysis and perform a blob segmentation.

A kind of offshore oil spill risk source fast intelligent identification classification method of high-resolution remote sensing image belongs to the cross technical field of marine remote sensing and artificial intelligence, including: 1) oil platform and ship as the main risk source of offshore oil spill, with the characteristics of pixel value higher than surrounding seawater, according to the method of pixel value gradient and spot detection, a squeezing excitation residual network combining Gaussian Laplacian operator and connected domain controller is designed;2) using prior knowledge such as pixel gradient and target size, suspected offshore oil spill risk source identification is carried out, the center position and size of the target are obtained using connected domain controller on multiple response points around the target, 3) finally, the suspected offshore oil spill risk source identified is classified by deep learning classification network.The present application only classifies the suspected offshore oil spill risk source target picture in the image, reduces the number of pictures, improves the running time, and the training set and the actual prediction picture size are consistent, only small-scale pictures are needed for training without using the whole remote sensing image, which is easier to train and saves time.

The invention discloses a dynamic mark point detection tracking method and system based on improved CSRT, and belongs to the technical field of bionic visual tactile sensors, and the method comprises the steps: determining a mark point, and obtaining a first frame image containing the mark point; acquiring an initial position of the mark point in the first frame image based on a morphological feature spot detection method; acquiring a CSRT tracker of the mark point based on the initial position, and acquiring motion information of the mark point through the CSRT tracker; obtaining vector modulus length of inter-frame displacement of the mark points based on the motion information, comparing the vector modulus length with a preset modulus length threshold value, and determining the mark points corresponding to the vector modulus length outside the modulus length threshold value as moving points; acquiring a region covering the moving points as a dynamic attention region based on the spatial distribution of the moving points; and performing displacement tracking of the contact object based on the dynamic region of interest. According to the morphological feature-based spot detection method, accurate initial positioning of the mark points is realized, and a reliable starting point is provided for subsequent tracking.

A spot detection device, a spot detection method, and a display device are provided. The spot detection device includes an imager and a spot detector, in which: the imager is configured to capture a display panel displaying a first image of a first gray scale and a second image of a second gray scale different from the first gray scale, and generate image data; and the spot detector is configured to receive the image data from the imager to detect a spot. The image data includes first image data obtained by capturing the first image, second image data obtained by capturing the second image, and a plurality of third image data obtained by capturing a portion of the first image.

The application provides a multi-scale glue point detection method, and belongs to the technical field of glue point detection; the method comprises the following steps: adopting an intelligent laser lens dispensing platform to collect images before laser chip bonding; establishing an LLAS data set based on the collected images; establishing a multi-level attention guide context aggregation network model; applying the multi-level attention guide context aggregation network model to process images in the LLAS data set, and segmenting out glue point targets; and obtaining glue point detection results based on the segmented glue point targets. The application proposes a multi-level attention guide context aggregation network (MACAN) based on a self-built LLAS data set. The purpose is to realize real-time and robust detection of glue points on laser chips, and to provide necessary quality assurance for realizing high coupling efficiency of high-power laser lenses.

A method of locating a defect in an e-coat on a surface can include acquiring an image of the surface. A correction coefficient can be applied to the image to form an adjusted image. The correction coefficient can relate pixel values of the image to a calibration value. The adjusted image can be separated into a spectral component which can be modified by a block average determination to create a modified spectral component. The spectral components can be compared with the modified spectral components to form a difference image. The difference image can be dilated and eroded. A region of interest can be identified from an image region using a blob detection. The defect can be classified as a defect type. The defect can be repaired or a coding parameter can be altered based on the defect.

The invention discloses a military wire test pre-judgment method combining a software technology and a camera module. The military wire test pre-judgment method comprises the following steps: acquiring a two-dimensional target image of a cross section of a target wire; processing the two-dimensional target image into a two-dimensional target cross section grayscale image by using an image graying method; preprocessing the image; classifying a target sub-region and a non-target sub-region in the two-dimensional target cross section grayscale image in the step S3 by using an SVM (Support Vector Machine) algorithm; using a spot detection algorithm to identify quasi-circular spots in the two-dimensional target cross section grayscale image; extracting the same local areas of the quasi-circular spots and the contour edge, and carrying out contour background subtraction so as to check and obtain effective quasi-circular spots; the circle center coordinates and the radius of the effective circle-like spots are recognized, and the Euclidean distance between the effective circle-like spots and the contour circle center is calculated through preset logic. According to the invention, the acquired image can be used for judging whether the conditions of core missing and non-uniform wire core distribution occur or not based on the acquired image.

The application discloses a method and device for double-target calibration of a shift camera, and the method comprises the following steps: acquiring calibration board images at different horizontal heights, wherein all the acquired calibration board images at least include a calibration board image at a horizontal height of Z=0; calling a Blob detection operator in opencv to extract feature point coordinates of all the calibration board images; constructing a shift camera model AH=B of the shift camera, and calculating a left camera internal parameter matrix M l and a right camera internal parameter matrix M r of the shift camera model AH=B; calling an opencv Zhang Zhengyou calibration function to solve distortion parameters of the calibration board image at Z=0; calling an opencv stereo calibration function to calculate a rotation matrix R and a translation vector T of the right camera relative to the left camera; calling an opencv stereo rectification function to calculate a re-projection matrix Q; inputting two-dimensional pixel coordinates and aligned disparity values d i to calculate 3D coordinates corresponding to the two-dimensional pixel coordinates; and the application can realize direct precision verification of double-target calibration of the shift camera and simplify a double-target calibration process of the shift camera.

The embodiment of the invention discloses a machine vision line patrol method and device based on spot detection, equipment and a medium. The method comprises the following steps: acquiring a machine vision track image, and setting a plurality of preset one-dimensional coordinate values in the machine vision track image; according to each preset one-dimensional coordinate value, carrying out image extraction on the machine vision track image to obtain machine vision track sub-images corresponding to each preset one-dimensional coordinate value; performing spot detection on each machine vision track sub-image to obtain key point coordinates of a plurality of spots; determining a real-time steering angle of a track in the machine vision track image according to the coordinates of the key points; and controlling the machine to move according to the track in the machine vision track image according to the real-time steering angle. According to the machine vision line patrol method based on spot detection, the trajectory identification calculation amount can be reduced, the identification real-time performance is improved, and the applicability is high.

The invention provides a picture consistency defect detection method and system for a projector, and relates to the technical field of picture detection, and the method comprises the steps: obtaining a detection image shot by the projector, then carrying out the image processing of the detection image, obtaining a plurality of decomposed images, and segmenting each decomposed image into a plurality of image blocks with the same size; sequentially performing peripheral brightness difference calculation on each image block to obtain a first judgment parameter, performing spot detection to obtain a second judgment parameter, and performing dead pixel detection to obtain a third judgment parameter; and judging whether the first judgment parameter, the second judgment parameter and the third judgment parameter are in the corresponding standard ranges or not, and indicating that the detection is passed. The method has the beneficial effects that through an automatic image processing technology, key parameters such as peripheral brightness difference, spots and dead pixels of a plurality of image blocks of each decomposed image can be accurately calculated, subjective deviation of manual observation is avoided, and accuracy and consistency of detection results are ensured.

The present invention discloses a flotation foam image analysis system based on pattern recognition, which relates to the field of flotation foam image analysis technology, including an image analysis terminal for controlling data transmission and information interaction between various modules. The present invention first removes noise and irrelevant areas in the foam flotation image, making subsequent foam density analysis and reduction analysis more convenient and accurate. Secondly, the foam in the final foam flotation image is counted by a spot detection algorithm to determine the number of foams, and then the foam density information is calculated. Then, the position of the foam is analyzed by a rectangular coordinate system to determine the foam reduction information. Finally, the foam reduction information and the foam density information are judged and analyzed to determine whether the flotation agent ratio needs to be changed. The above method avoids the problem that workers may become tired from staring at the flotation tank for a long time, which may increase the judgment error of the flotation foam and make it impossible to accurately control the flotation agent ratio.

The present invention discloses a quality detection method for an electric grid for protecting wild animals. The method belongs to the field of computer vision target detection. The present invention includes image preprocessing to obtain a target image set DST; performing BLOB detection on each image DST in the image set DST respectively, and further removing interference points to obtain a set of intersection points POINTS of the electric grid; center calibration to obtain a set of points CaliPoints; performing glue overflow judgment based on the point sets POINTS and CaliPoints before and after calibration; column separation to obtain a column set ClasPoints after separation; and using a bean shortage recognition algorithm to perform bean shortage judgment based on the target image set DST and the column set ClasPoints. The quality detection method for electric grids for protecting wild animals greatly shortens the duration of the complete process of making electric grids, greatly reduces labor costs, and improves the efficiency of enterprise net making.

This application provides a spot detection control method, device, and spot detection system, which are applied to spot detection control equipment. First, according to the change information between the laser spot sizes corresponding to different positions on the main optical axis when the laser emitted by the laser is transmitted to these positions, the spot analyzer is controlled to move to the focal position on the main optical axis; then, the spot detection result of the laser is determined based on the first spot information collected by the spot analyzer at the focal position. The first spot information includes at least one of spot size, spot shape, and spot intensity data. This realizes the automatic real-time movement adjustment of the spot analyzer, ensures the accuracy of movement adjustment, improves the efficiency of moving the spot analyzer to the focal position, and thus enhances the detection efficiency of the detection result.

A non-detected workpiece may occur in workpiece detection processing for detecting, on the basis of image data captured by a visual sensor, a workpiece depicted in the image data.　An image processing device 50 comprises: a workpiece detection unit 52 that executes workpiece detection processing for detecting, on the basis of image data captured by a visual sensor 14, a workpiece depicted in the image data; a pixel removal unit 54 that removes, from the image data, pixels that depict the workpiece detected in the workpiece detection processing; a blob detection unit 56 that detects, as a blob, an aggregate of pixels that depict a subject in the image data from which pixels have been removed by the pixel removal unit 54; and a non-detection determination unit 58 that determines, on the basis of the blob detected by the blob detection unit 56, whether a workpiece not detected in the workpiece detection processing is present.

Methods, system, and media for comparing a set of images to determine the existence and location of any differences between the image set. The differences may be located using image comparison techniques such as SURF and Blob Detection, as well as through techniques used to identify areas of data sliding and match probabilities. A logical match probability, as well as a physical match probability, may be included in an output report with a result image highlighting the differences between the comparison images in the image set.

This invention provides a system and method that efficiently detects objects imaged using a 3D camera arrangement by referencing a cylindrical or spherical surface represented by a point cloud, and measures variant features of an extracted object including volume, height, and center of mass, bounding box, and other relevant metrics. The system and method, advantageously, operates directly on unorganized and un-ordered points, requiring neither a mesh / surface reconstruction nor voxel grid representation of object surfaces in a point cloud. Based upon a cylinder / sphere reference model, an acquired 3D point cloud is flattened. Object (blob) detection is carried out in the flattened 3D space, and objects are converted back to the 3D space to compute the features, which can include regions that differ from the regular shape of the cylinder / sphere. Downstream utilization devices and / or processes, such as part reject mechanism and / or robot manipulators can act on the identified feature data.

The present invention relates to the technical field of mask production, and in particular to a mask defect detection device and method, including a device cabinet, on which are respectively installed: a feeding mechanism, providing masks to be detected; a three-stage belt line conveying mechanism, conveying masks to be detected; an image acquisition mechanism, collecting images of masks to be detected; image processing software, judging the collected mask images according to preset mask standards and detection algorithms and giving good or bad results; a rejection mechanism, rejecting bad products given by the image processing software; a control system, for electrically connecting a feeding mechanism, a conveying mechanism, an image acquisition mechanism, an image processing software and a rejection mechanism. The mask defect detection method provided by the present invention combines a traditional BLOB detection method with a deep learning method to complete the detection of all items to be detected, and the detection method has high stability, high detection efficiency and high detection accuracy, so that the quality of the product after the device detection is guaranteed.

Methods, systems, and media for validating objects within a container, such that a validation system confirms that objects within a container are valid i.e., ready to be shipped from a data center for reuse or ready to be destroyed. The validation system can be used to identify objects, e.g., media, disks, etc., and determine if valid objects are at the correct processing location to be shipped or destroyed. The methods may include identifying the objects, locating an identifier on the object, comparing the identifiers to known identifiers and determining, based on the comparison, if valid objects are at the correct processing location to be shipped or destroyed. The validation system may also use image processing techniques, such as blob detection, to identify invalid objects in the container. The system may reject the entire container if invalid objects are detected.

A method and a tracking managing subsystem (111) for providing blobs relating to tracking of a sports projectile are disclosed. The method comprises obtaining (A110), from the tracking assembly (115), a tracking stream including tracking data maps, which are associated with timestamps, wherein each tracking data map of the tracking data maps is associated with a respective timestamp of the timestamps, and obtaining (A120) a set of transform parameters. Furthermore, the method comprises deriving (A130) a set of blobs by applying a blob detection procedure to at least two tracking data maps of the tracking stream, and sending (A140) the set of blobs and the set of transform parameters.

The invention discloses a flotation froth image analysis system based on pattern recognition, relates to the technical field of flotation froth image analysis, and comprises an image analysis terminal used for controlling data transmission and information interaction among modules. According to the method, noise and irrelevant areas in the froth flotation image are removed firstly, density analysis and decrement analysis of follow-up froth are more convenient and accurate, then froth in the final froth flotation image is counted through a spot detection algorithm, the number of the froth is determined, then froth density information is calculated, and then the froth density information is calculated. The method comprises the steps that firstly, the position of foam is analyzed through a rectangular coordinate system, foam reduction amount information is determined, finally, the foam reduction amount information and foam density information are judged and analyzed, and whether the flotation reagent proportion needs to be changed or not is determined. And therefore, the judgment error of flotation froth can be increased, and the proportion of flotation reagents cannot be accurately regulated and controlled.

The invention provides an intelligent vision inspection method, system and device running on an embedded platform, and relates to the field of computer vision measurement. The method comprises the following steps: simultaneously acquiring two paths of images through an image sensor; performing AI target detection on one path of image to obtain a target frame and mapping the target frame to a display coordinate system; calculating the offset between the center of the target and the center of the picture to judge a visual alignment state; monocular distance estimation and smoothing processing are carried out based on the target pixel height, the physical height and the equivalent focal length; executing color threshold blob detection in the ROI region defined by the target frame to identify abnormity; a double-layer duplicate removal mechanism is adopted to avoid duplicate records; and performing automatic screenshot on a new abnormal event, recording a log, and overlapping and displaying detection information on a picture. According to the invention, in the embedded device with limited resources, the monocular camera is combined to realize low-cost and high-reliability intelligent inspection, and the problems of unstable distance measurement, abnormal false detection and repeated event recording are effectively solved.

Methods, system, and media for comparing a set of images to determine the existence and location of any differences between the image set. The differences may be located using image comparison techniques such as SURF and Blob Detection, as well as through techniques used to identify areas of data sliding and match probabilities. A logical match probability, as well as a physical match probability, may be included in an output report with a result image highlighting the differences between the comparison images in the image set.

A method of locating a defect in an e-coat on a surface can include acquiring an image of the surface. A correction coefficient can be applied to the image to form an adjusted image. The correction coefficient can relate pixel values of the image to a calibration value. The adjusted image can be separated into a spectral component which can be modified by a block average determination to create a modified spectral component. The spectral components can be compared with the modified spectral components to form a difference image. The difference image can be dilated and eroded. A region of interest can be identified from an image region using a blob detection. The defect can be classified as a defect type. The defect can be repaired or a coding parameter can be altered based on the defect.

The present application provides a cell detection method, device, equipment and storage medium based on pathological images, wherein the method comprises: obtaining a number of annotated pathological images; using each pathological image to train and evaluate a pre-built cell detection model to obtain a qualified cell detection model; using the qualified cell detection model to obtain a center point expansion image of the pathological image to be processed; performing spot detection processing on the center point expansion image to obtain the center point pixel position of each center point expansion area in the center point expansion image; counting the number of center point pixel positions in each center point expansion area group to obtain the number of cells under each cell category in the pathological image to be processed; the above method is conducive to reducing the manpower burden while improving the positioning efficiency and counting efficiency of different types of cells.