118results about How to "Improve defect detection accuracy" patented technology

The invention discloses an oil and gas pipeline intelligent internal detection device based on multi-module combined location. The device comprises an internal detector arranged inside an oil and gas pipeline and a floor marker arranged outside the oil and gas pipeline. The internal detector comprises a driving system, a magnetic leakage detecting system, a speed control system, a distance measuring system and a low frequency emitting system, wherein the driving system, the magnetic leakage detection system, the speed control system, the distance measuring system and the low frequency emitting system are relatively fixedly connected together; the floor marker at least comprises a low frequency receiving system; the low frequency emitting system is used for sending detection results obtained by the magnetic leakage detection system out of the oil and gas pipeline; the low frequency receiving system is used for receiving low frequency electromagnetic pulse signals transmitted by the low frequency emitting system. The oil and gas pipeline intelligent internal detection device based on the multi-module combined location can timely obtain the detection data of the pipeline detector and analyze the detection data and accordingly rapidly determine the defect conditions of the pipeline.

The invention provides a method for detecting surface blemish, which comprises the following steps: acquiring an image of an object to be detected; generating a binary image through binary treatment according to the acquired image of the object to be detected; filtering possible blemish particles for the image subjected to the binary treatment; and comparing grey mean value of a blemish with the grey mean value of an image blemish area before the binary treatment to determine the blemish. The invention also provides a device for detecting surface blemish. The device has better practicability and accuracy, and is easy to popularize.

The invention provides a display screen quality detecting method and device, electronic equipment and a storage medium. The method comprises the steps of receiving a quality detecting request from a console deployed on a display screen production line, wherein the quality detecting request includes a display screen image acquired by an image acquisition device on the display screen production line; inputting the display screen image into a defect detection model to obtain a defect detection result, wherein the defect detection model is obtained by training a historical defect display screen image by a depth convolution neural network structure and an object detection algorithm; and according to the defect detection result, determining defects on the display screen corresponding to the defect image, the defect category corresponding to the defects, and the locations corresponding to the defects. The technical scheme has the advantages of high defect detection accuracy, good system performance and high service expansion ability.

The invention relates to an orientation identification method of ultrasonic array crack defects based on vector full-focusing imaging, and belongs to the field of nondestructive testing. According to the method, vector full-focusing treatment is performed on signals received by a transducer array so as to obtain a global vector diagram, the positions of the defects are determined by the strongest energy of reflected signals at the defects, then local vector diagrams of the defects are extracted according to the positions of the defects, and finally the directions of the vectors are determined by utilizing the directions of vectors in the local vector diagrams. According to the principle of vector full-focusing imaging, a plurality of sub-arrays are constructed in the transducer array, the unit direction vector at any imaging point is calculated through the sub-arrays, and echo amplitude after compensation at any imaging point to the transducer array is vectorized by utilizing the unit direction vector, so that amplitude vector at any imaging point is obtained.

To realize reliable on-film/under-film defect classification (classification into 3 classes of on-film, under-film, and determination-disabled) with determination propriety determination, an on-film/under-film defect classification method and a method of narrowing a range of defect generation timing are given, the methods being robust to 4 variation factors, wherein an edge of a boundary line between a line pattern region and a base region is focused, and whether the edge is preserved between defect and reference images in a defective region is determined, thereby an on-film or under-film defect can be identified. Furthermore, a range of the defect generation timing can be narrowed based on an identification result of the on-film or under-film defect, and information of a defect classification class (defect type) such as particle defect or pattern defect as necessary.

The invention discloses a defect detection method and a system of a defective image. Through a depth learning method, a depth convolution neural network model is constructed to analyze the deep logical relationship between defect factors in the bad image, the defect relationship between parts and parts, and the defect classification between parts and the whole, thereby realizing defect classification. The invention can effectively realize defect position detection and defect class prediction, has high precision and accurate classification, and has better detection effect on small targets. It can optimize the detection accuracy and reduce the missed detection.

The invention relates to a method of defect detection, device and equipment. The method comprises steps: in the case of detecting whether a surface defect exists in an electronic product included in aspecified image, a first well-trained neural network model is used to classify the specified image, wherein the first neural network model is used for detecting whether the image belongs to a problematic image or a normal image; if the first neural network model classifies the specified image to the problematic image and the given reliability value of the specific image belonging to the problematic image is smaller than a reliability threshold, a second well-trained neural network model is used to classify the specified image, wherein the second neural network model is used for detecting whether the image belongs to a problematic image or a normal image; and if the second neural network model classifies the specified image to the problematic image, the surface defect is determined to exist in the electronic product included in the specified image. The method, the device and the equipment can improve the defect detection accuracy in a condition of consuming no manpower basically.

The invention discloses an electric power inspection image defect identification method. The method comprises the following steps: creating and cascading a target detection network and a plurality ofclassification networks; obtaining a plurality of frames of inspection image samples, labeling targets in the inspection image samples, and generating a training sample set; adopting a training sampleset to train a cascade network, wherein the quantization parameter of each network layer is related to the quantization stage number and the quantization range of the network layer where the networklayer is located; and identifying defects in the newly acquired inspection image by adopting the trained cascade network. According to the invention, an effective FPGA airborne identification system is provided for operation and maintenance of a power grid tower and an overhead line, and a corresponding quantization function can ensure that different channels of different network layers can be properly quantified, so that the precision of the network is reserved to the maximum extent; through cascading the target detection network and the classification network, the defect detection accuracy is greatly improved, and the unmanned aerial vehicle routing inspection of the power grid truly realizes automatic identification.

The invention discloses a crown cap surface defect online detection method based on image processing, and belongs to the technical field of image processing. The method comprises the following steps:acquiring a crown cap inner surface image, and implementing central positioning and region extraction by using an image segmentation technology; implementing bottle cap inner surface defect detectionby using image processing algorithms such as threshold segmentation, morphological processing, connected region feature analysis and region comparison; establishing a crown cap outer surface pattern template offline by adopting an automatic training learning algorithm; and acquiring a crown cap outer surface image, and finishing bottle cap outer surface defect detection by using feature template matching, affine transformation and regional comparison. The method can be used for rapidly detecting various crown cap surface defects online, can detect 300 bottle caps in one minute, has the advantages of high defect detection accuracy, high robustness and the like, and can be attached to a crown cap production line to realize defect detection and type differentiation of crown caps.

The invention discloses a sealing ring surface defect detection method based on machine vision. The sealing ring surface defect detection method comprises the following steps: firstly, acquiring images of the surface of a sealing ring, and performing self-adaptive median filtering treatment on the acquired images; subsequently, calculating a gray level gradient and a vertical gradient of the images, and extracting gray bevel structures in the images according to the gray level gradient and the vertical gradient; partitioning different bevel characteristic point neighborhoods, and calculating a gray level mean of the partitioned neighborhoods; finally, by taking functions for describing the gray level difference degree of the partitioned neighborhoods as defect judgment principles, screening out defect outline points, and detecting the defects of the surface of the sealing ring. According to the forming reason of the defects of the surface of the sealing ring, inherent differences of defect outlines and appearance outline of the sealing ring can be analyzed and verified on the images, various types of defects, including recesses, rill marks, impurities, trimming and over-cutting defects, on the surface of the sealing ring can be detected, and the method has the advantages of high defect detection accuracy, good algorithm robustness and the like.

In order to detect defects in display panels by an automatic way accurately and quickly, the present invention proposes a method combining image feature extraction and classifier model. It calculates the histograms of oriented gradient (HOG) of images of the display panel collected by an industrial camera of a detection apparatus as feature vectors. Then use them as input samples to train the classifier model to recognize the defects of the display panel.

The invention discloses a metal workpiece surface defect image detection method. The method comprises the steps of: firstly, collecting a surface image of a metal workpiece through professional imaging equipment, and then enabling the collected image to be subjected to image early-stage preprocessing including the steps of uneven illumination image gray scale correction, image filtering, image threshold segmentation and the like; further performing feature extraction and analysis on the preprocessed image, and introducing a sub-pixel edge detection algorithm to perform edge detection on the metal workpiece; and finally, carrying out template matching on the template image and the measurement image by adopting a gray-level co-occurrence matrix algorithm so as to carry out defect detection on the surface of the metal workpiece. According to the metal workpiece surface defect image detection method provided by the invention, the traditional manual visual inspection is replaced by an automatic detection technology based on machine vision, the production efficiency is improved, the labor cost is reduced, the metal workpiece defect detection precision is improved by adopting a sub-pixeledge detection algorithm, and the production quality is optimized.

The invention discloses a part defect detection method and device, a medium and a computer program product. The part defect detection method comprises the steps of obtaining a to-be-detected image corresponding to a to-be-detected part, carrying out the defect prediction of the to-be-detected image based on a full-scale gray scale prior depth segmentation model, and obtaining an image defect detection result. The full-scale gray scale prior depth segmentation model is obtained by performing iterative training optimization on a to-be-trained defect detection model formed by cascading a preset number of deep neural network modules based on a pre-collected training defect image set. The technical problem that the defect detection accuracy of parts is low is solved.

In order to solve the problems of imperfect surface defect detection information, low precision, low efficiency and the like, the invention provides a defect detection method based on a polarization structured light imaging technology and an improved Mask RCNN. The method comprises the following steps: firstly, combining polarization processing with structured light three-dimensional imaging to obtain a high-definition two-dimensional physical graph and three-dimensional space information of an object; performing median filtering processing on the two-dimensional physical graph; secondly, on the basis of a Mask RCNN target recognition method, adding a K-means algorithm to carry out clustering analysis on a training set, adding branches with side edge connection from top to bottom to an original FPN structure, and combining lower-layer high-resolution features and upper-layer high-resolution features to generate a new feature map; detecting an image with defects by utilizing the improved Mask RCNN network, and classifying, positioning and segmenting the defects; finally, obtaining a series of information such as the type, position, length, width, depth and area of the defect throughdata arrangement, achieving quantification of defect data, and the object surface defect detection precision and efficiency are effectively improved.

The invention discloses a background noise removal device and a background noise removal method for magnetic flux leakage detection and calibration. The device comprises a detection and calibration module, a sampling control module and an effective signal extracting and processing module, wherein the sampling control module is electrically connected with the detection and calibration module and transmits a detection signal to the effective signal extracting and processing module; and the effective signal extracting and processing module is used for preprocessing the collected signal, rejecting gap interference data and separating effective data from original data so as to guarantee that the signal entering a signal processing system is an effective signal in which a periodic interference signal is rejected. According to the background noise removal device, a background magnetic field in a calibration roll used in a flux leakage detection signal is eliminated, noise signals are removed, and effective flux leakage detection signals are extracted, so that the signal to noise ratio of internal defect magnetic flux leakage can be greatly increased, and the defect detection accuracy rate can be increased.

The invention relates to a metal paint spraying surface defect detection method. The method comprises the steps of acquiring a metal paint spraying surface image data set containing defect images anddefect-free images; carrying out primary selection on the metal paint spraying surface image data set by adopting a binary classification mode to obtain a positive sample with a label, a positive sample without a label and a negative sample without a label; acquiring an image containing unknown types of defects on the surface of metal paint spraying, and performing training test on a deep learningneural network in combination with a sample obtained by binary classification primary selection to obtain a defect detection model; and inputting the actual metal paint spraying surface image into the defect detection model, and outputting a defect detection result of the actual metal paint spraying surface image. Compared with the prior art, the method has the advantages that early-stage samplescreening and labeling can be accurately and quickly carried out by combining blob block detection and the deep learning neural network, and meanwhile, the neural network is trained by utilizing the defect images of unknown types, so that the metal paint spraying surface defects can be quickly, accurately and comprehensively detected by the method.

The invention belongs to the related technical field of defect detection and discloses a spaceflight electronic welding spot defect detection method based on an improved Tiny-YOLOv3 network. The detection method comprises the following steps of (1) enhancing a network layer for feature extraction in Tiny_YOLOv3 by using a Mobileet network, replacing seven convolution and maximum pooling network layers in a Tiny _ YOLOv3 backbone network by using a lightweight network Mobileet to obtain an improved Tiny_YOLOv3 network; (2) inputting a welding spot infrared image with a known defect type into the improved Tiny_YOLOv3 network as a training data set of a sample so as to train and learn the improved Tiny_YOLOv3 network, thereby obtaining an improved Tiny _ YOLOv3 network model; and (3) inputting the infrared image of the welding spot sample to be detected into the improved Tiny_YOLOv3 network model so as to complete the detection of the welding spot defects. The spaceflight electronic welding spot defect detection method effectively improves spaceflight electronic welding spot defect detection accuracy.

The invention relates to the field of data processing, and provides an image processing method and device based on semiconductor defect detection, equipment and a medium. The method comprises the steps: acquiring a semiconductor full-band internal structure image set, and selecting images with different frequency bands from the internal structure image set to serve as source images; processing thesource image to obtain a high-pass image block and a low-pass image block, converting the low-pass image block to obtain a low-pass vector, encoding the low-pass vector to obtain corresponding sparsecoefficient vectors, and performing fusion processing on all the sparse coefficient vectors to obtain a low-pass fusion vector; converting the high-pass image block to obtain a high-pass vector, constructing a vector set of each source image according to the high-pass vector and the low-pass vector, and performing fusion processing on all the vector sets to obtain an all-pass fusion vector; reconstructing the low-pass fusion vector and the all-pass fusion vector to obtain a target vector, and converting the target vector into a target image. The definition of semiconductor imaging can be improved, and therefore the semiconductor defect detection precision is improved.

The invention relates to a defect detection method and device, electronic equipment and a computer readable storage medium. The method comprises the following steps: carrying out graying processing on a to-be-detected image to obtain a grayscale image; determining gray scale information of a background in the gray scale image and gray scale information of a target object in the gray scale image according to the gray scale value of each pixel point in the gray scale image; performing ternary processing on the grayscale image according to the grayscale information of the background and the grayscale information of the target object; and determining defect information in the to-be-detected image according to a ternary processing result. According to the embodiment of the invention, the defect detection precision of the image can be effectively improved.

The invention discloses an image grey value based mask optical defect detecting method. The method includes steps: acquiring bright area data and dark area data of an actual image on a camera interface; adopting a camera flat field correction function for camera view field brightness correction of the actual image to guarantee value maintenance of grey values of a bright area and a dark area in aview field, and keeping uniform as far as possible; selecting a dark area on a calibration plate of the actual image, and calibrating a grey value to be V1; selecting a bright area on the calibrationplate of the actual image, and calibrating a grey value to be V2; recording the calibrated grey value V1 of the dark area and the calibrated grey value V2 of the bright area into a Recipe template toserve as subsequent mask detection parameters, and directly applying the subsequent mask detection parameters as grey values of standard image generation bitmap to keep uniformity of the grey values of the actual image and the grey values of a standard image as far as possible; after image registration, subjecting the actual image and the standard image to absolute subtraction operation to furtherjudge whether the actual image has defects or not. By adoption of the method, defect detection accuracy can be improved, and great detection effects are achieved.

A defect detection apparatus including a region setter which sets a region of interest in an image including a plurality of pixels, and a defect detector which divides the region of interest into a plurality of segments and generates a trend line for each of the segments based on gray level values of the pixels. Here, the defect detector detects a stain in the image by comparing the gray level values of the pixels with the trend line.

The embodiment of the invention discloses an integrated circuit defect optical detection method and an integrated circuit defect optical detection device. The method comprises the steps that: a spiral phase plate is arranged at a Fourier plane; through the spiral phase plate, scattered light and reflected light of the defect on the integrated circuit are received, wherein the scattered light passes through the spiral phase plate and forms a circle image on an image receiving plane, and the reflected light passes through the center of the spiral phase plate and forms same-phase uniform background light on the image receiving plane; the circle image and the same-phase uniform background light are subjected to phase interference on the image receiving plane, such that a spot image comprising bright spots and dark spots is formed; and the type of defect on the integrated circuit is determined according to the spot image. According to the embodiment of the invention, integrated circuit defect detection precision can be improved.

The invention relates to the field of artificial intelligence, and discloses a defect detection method, device and equipment for a mobile phone touch screen and a storage medium, which are used for improving the defect detection accuracy of the flexible mobile phone touch screen. The method comprises the following steps: performing defect information labeling on a plurality of training images to generate a sample image data set; generating an enhanced image data set according to the sample image data set; performing model training on the training model according to the enhanced image data set to obtain a mobile phone touch screen defect detection model; the method comprises the following steps: acquiring an initial folding image and an initial stretching image of a to-be-detected flexible mobile phone touch screen, and carrying out image preprocessing on the initial folding image and the initial stretching image to obtain a target folding image and a target stretching image; and inputting the target folding image and the target stretching image into a mobile phone touch screen defect detection model for crease detection to obtain a crease detection result.

A system and method for detecting defects in TFT-array panels is provided that improves defect detection accuracy by adjusting the thresholding parameters used to classify defective pixels when the number of defects reported by a TFT-array testing system exceeds a predetermined critical number. In a preferred embodiment, the thresholding parameters are adjusted until the number of reported defects is less than or equal to the predetermined critical number. The predetermined critical number represents a threshold number for determining if the number of reported defects is abnormally high. Reducing the number of reported defects to a number equal to or less than the predetermined critical number will decrease the operation time of the TFT-array repair equipment, because of the reduced number of potential defects it will be required to handle, and will also result in the TFT-array testing system reporting a smaller number of potential defects, with the potential defects that are reported having a higher probability of being real defects.