132 results about "Sobel operator" patented technology

The invention discloses a ginkgo leaf extracting solution state real-time monitoring method based on image processing, and relates to the technical field of ginkgo leaf extracting solutions. The method comprises the following steps: constructing a multi-light-source imaging environment to shoot a ginkgo leaf extracting solution image; obtaining an extracting solution mask through a U-net segmentation model, and obtaining an extracting solution foreground image in combination with the ginkgo leaf extracting solution image; segmenting the extracting solution foreground image through an adaptive threshold method to obtain an extracting solution block graph, and optimizing through a watershed boundary optimization method to obtain an optimized extracting solution block graph; a block boundary probability value is calculated through a boundary probability model with double distance changes, a block gradient magnitude is calculated through a Sobel operator, block boundary confidence is obtained by combining the block boundary probability value and the block gradient magnitude, and block boundary pixels are determined; and collecting a boundary gradient feature vector sequence of the block boundary pixels, inputting the boundary gradient feature vector sequence into the extracting solution evolution trend model, outputting to obtain an oxidation risk index, and performing early warning if the oxidation risk index is greater than a preset threshold value.

The invention relates to the technical field of three-dimensional modeling, in particular to a three-dimensional modeling system and method based on a laser radar, and the system comprises a point cloud boundary extraction module, a boundary error analysis module, a self-adaptive cutting repair module and a stitching structure optimization module. According to the method, three-dimensional coordinates, normal vectors, reflection intensity, distances and indexes of boundary points are called layer by layer, a Sobel operator and a Canny algorithm are combined to cooperatively recognize a boundary and an error region, then local region repair is completed based on least square plane fitting and spatial residual analysis, a tension vector is analyzed through dynamic planning, and stitching path adjustment is executed. On the cooperative basis of multi-type point cloud features, a spatial topological structure and boundary continuity dynamic modeling, the problems of local damage, connection errors and instable stitching paths of point cloud boundaries in complex scenes are solved, and the integrity of boundary structures, the coherence of stitching transition and the rationality of spatial geometric expression are remarkably improved.

The invention discloses a geotechnical engineering slope deformation monitoring method and system. The method comprises the following steps: acquiring multi-modal image data, enhancing the weak deformation area contrast of the multi-modal image data by using a CLAHE algorithm, extracting gradient edge information through a Sobel operator, and filtering laser radar point cloud noise based on VMD; capturing a bidirectional dependency relationship of a time sequence based on a BiLSTM bidirectional long short-term memory network, and dynamically allocating weights of different time steps and spatial positions through an attention mechanism; optimizing hyper-parameters of the hybrid prediction model by using a GJO Golden Litsea Optimization algorithm to obtain a target hybrid prediction model; and inputting the feature image data into the target hybrid prediction model for prediction, generating a risk probability thermodynamic diagram, and positioning the position of the potential slip crack surface of the slope according to the risk probability thermodynamic diagram. And reliability and accuracy of geotechnical engineering slope image data analysis are improved.

The invention relates to the technical field of medical image analysis, computer vision and deep learning, in particular to a texture perception and multi-scale feature adaptive fusion hip joint osteophyte detection method. According to the method, firstly, a bone texture feature extraction module is used for carrying out feature extraction and enhancement on a hip joint X-ray image, a parallel double-branch structure is adopted, gradient and space structure features are extracted through a Sobel operator and pooling operation, and feature maps are fused; then, the fused features are input into a double-backbone network, the first backbone network extracts local fine textures and long-range dependence features by combining convolution, self-attention and a gating mechanism, and the second backbone network extracts multi-scale semantic features through hierarchical grouping and depth separable convolution; and double-trunk output is dynamically weighted and fused through an adaptive fusion module, features are optimized through a multi-scale semantic fusion module, and finally the position and confidence of osteophyte are output.

The invention discloses a wide-angle image acquisition and processing system for underwater target identification, which relates to the technical field of image processing and target detection, and comprises an image acquisition and preprocessing module for acquiring an underwater image, removing initial noise through a preset filter, processing an enhanced grayscale image by adopting image smooth combination, and obtaining an underwater image; the pixel density gradient calculation module is used for calculating a multi-direction pixel density gradient by adopting a horizontal direction convolution kernel and a vertical direction convolution kernel of a Sobel operator according to the enhanced grayscale image, and obtaining a gradient intensity graph through gradient amplitude calculation; according to the wide-angle image acquisition and processing system for underwater target recognition, through multi-level feature fusion and classification positioning, the detection precision and positioning accuracy of an abnormal region in an underwater complex environment are remarkably improved, and efficient technical support is provided for underwater target recognition.

The invention discloses a reversible information hiding method and system for enhancing image smoothness, and relates to the technical field of information security, and the system comprises a preprocessing module, an information embedding module and an optimization recovery module. According to the method, the Sobel operator is utilized to calculate the image gradient map, the region is divided based on the threshold T, the smooth region and the texture region can be accurately distinguished, an accurate basis is provided for differential processing of different regions, region characteristics are fitted, large-size blocks and sub-blocks of the smooth region are divided to facilitate subsequent refinement processing based on the gradient mean value, and the processing efficiency is improved. The small-size blocks of the texture area are beneficial to operation aiming at texture features, the rationality and effectiveness of overall processing are improved, meanwhile, the pixel mean value in the large-size blocks of the smooth area is calculated, the pixel and mean value difference value is recorded, reversible low-pass filtering is carried out, and the reversibility of the image in the whole information hiding and recovering process is ensured.

The invention relates to the technical field of image processing, in particular to a lens edge detection method and system based on image generation, and the method comprises the following steps: collecting a lens image through an industrial camera, carrying out the weighted graying, extracting an edge coordinate through a Laplace operator, constructing an edge pixel coordinate set, and carrying out the image processing; screening an edge validity marking result by combining a gray difference threshold, calculating a phase difference between a gradient direction and light source incidence based on a Sobel operator, extracting pixels in a consistent direction, carrying out spatial clustering, generating a lens edge pixel cluster, and fitting a continuous curve to construct a complete lens edge contour. Noise and reflection interference are restrained through pixel neighborhood gray difference, effective pixels are screened in combination with a gradient direction and light source incident phase relation, edge direction consistency is enhanced, pixels are clustered according to spatial continuity and gradient intensity, contours are continuously fitted based on pixel cluster distribution, and lens edge integrity and geometric consistency under complex illumination are improved.

The invention discloses a corn seed environment adaptability cultivation method based on edge calculation, which comprises the following steps: step 1, acquiring environment factor data of a corn seed sowing area, and preprocessing the environment factor data; 2, extracting the edge contour of the corn seedling by adopting Canny edge detection and a Sobel operator based on the RGB image; 3, constructing a multi-source heterogeneous data sample set according to time and space coordinate alignment; 4, constructing a causal structure diagram by adopting an improved NOTEARS algorithm; 5, carrying out comprehensive sorting to obtain a key factor path; step 6, inputting the multi-dimensional environment factor vector of the target area to the starting node of the key factor path to obtain a predicted phenotypic feature vector; and 7, comparing the actual phenotypic feature vector with the predicted phenotypic feature vector, and counting a prediction error. According to the method, causal modeling and image analysis are fused, and the breeding efficiency and the accuracy of regional suitability judgment are improved.

The invention discloses a road underground defect AI intelligent identification method based on a three-dimensional radar image, and the method comprises the steps: setting electromagnetic wave parameters of ground penetrating radar equipment, and carrying out the multi-view image collection; the gray gradient change rate of the C-Scan image is calculated through a Sobel operator, the road state is identified, and a road disease area corresponding to the abnormal color plaque is extracted; constructing an advancing direction road defect type identification model, and identifying the road defect type of the road disease area; constructing a depth direction road defect type identification model, and identifying the road defect type of the road disease area; and according to the road defect type identification accuracy of the model, performing model optimization in combination with a preset accuracy threshold. According to the method, the problems of single parameter setting, high image interference misjudgment rate, insufficient defect identification precision and the like in traditional ground penetrating radar detection are effectively solved, and high-precision, multi-dimensional and intelligent identification of road underground structure diseases is realized.

The invention discloses a long-distance binocular camera calibration optimization method based on multistage feature enhancement, and the method comprises the steps: carrying out the processing of a condition that a long-distance calibration plate has a highlight region and is fuzzy in angular points, employing a bilateral filter to suppress the reflection of a highlight mirror surface, reducing the reflection of light, and maintaining the definition of an image; the checkerboard texture is enhanced by adopting a CLAHE algorithm in combination with blocking processing and a contrast gain threshold value; a Sobel operator and a Harris corner response function are fused, gradient direction distribution characteristics of a pixel neighborhood are extracted through the Sobel operator, weighted fusion is carried out on the gradient direction distribution characteristics and the Harris corner response function, and the response intensity and specificity of a corner area are remarkably enhanced. A three-level image preprocessing framework including reflection suppression, contrast enhancement and corner enhancement is constructed, the problem of feature extraction of a traditional method under a complex illumination condition is solved, the corner area detection capability is enhanced, the problems of specular reflection noise, low contrast and corner blur are effectively solved, and the calibration precision is improved.

The embodiment of the invention provides a steel rail corrugation recognition method and system and a storage medium, and belongs to the technical field of railway track detection. The steel rail corrugation identification method comprises the following steps: sending preprocessed steel rail corrugation true value image data into an image segmentation identification model to extract global semantic features and local texture features; performing channel averaging on the output of the intermediate feature extraction stage, generating edge features through a Sobel operator, and fusing the global semantic features, the local texture features and the edge features to obtain a segmented mask graph; and inputting the ROI image segments cut based on the segmentation mask image and the features extracted in the intermediate feature extraction stage and used for target detection into a target detection network together, executing regression calculation based on bounding box overlapping degree optimization loss, and outputting a rail corrugation detection result. According to the scheme of the invention, through multi-scale feature extraction and channel space attention fusion, the positioning precision and segmentation robustness of the rail corrugation region are effectively improved.

The invention discloses a crack sub-pixel precision edge detection method based on an improved Canny operator. The crack sub-pixel precision edge detection method comprises the steps of obtaining a crack image corresponding to a structural body surface crack; the crack image is preprocessed; carrying out gradient calculation on the preprocessed crack image through an improved Sobel operator to obtain a gradient image containing a gradient magnitude and a gradient direction; performing non-maximum suppression on the gradient image, adaptively obtaining an optimal threshold by adopting an improved maximum between-class variance method, judging a crack edge according to the optimal threshold, and further obtaining a coarse edge pixel-level image; processing edge information in the coarse edge pixel-level image by using a Zernike orthogonal moment to obtain coordinates of sub-pixel edge points; and connecting all the sub-pixel edge points, and outputting a crack detection result image. On the basis, the crack edge detection precision is improved to a sub-pixel level.

The invention relates to a lightweight traffic road semantic segmentation method based on multi-scale feature fusion, and belongs to the field of intelligent driving. The method solves the problems that precision and speed are difficult to balance and small target recognition accuracy is low in real-time application of a traditional semantic segmentation method. According to the technical scheme, context information is extracted by adopting a Vision Transform branch, space information is extracted by adopting a CNN branch, a model is optimized through feature fusion and alignment loss during training, and only the CNN branch is used during reasoning; sobel operator edge detection, a multi-scale feature fusion module and a feature distribution simulation module are integrated. The method has the technical effects that high-precision semantic segmentation is realized, the small target recognition capability is remarkably improved, and meanwhile, the real-time processing speed is ensured.

The invention relates to a reconstructed image anti-aliasing processing method based on an adaptive sampling rate, belongs to the technical field of CT (Computed Tomography) reconstructed image processing, and solves the problem that the existing anti-aliasing processing method for a CT reconstructed image is poor in effect. The anti-aliasing processing method for the reconstructed image comprises the following steps: preprocessing the reconstructed image to obtain a to-be-processed image; performing edge extraction on the to-be-processed image based on a preset gradient threshold and a preset Sobel operator, and segmenting the to-be-processed image into a to-be-replaced image region and a reserved image region; determining an adaptive sampling rate based on the horizontal direction edge pixel connection number and the vertical direction edge pixel connection number of all sawtooth vertex pixels in the to-be-processed image; determining a pixel value of each pixel in the to-be-replaced image region based on the adaptive sampling rate and the to-be-processed image, and obtaining a replaced image region; and combining the replaced image area and the reserved image area to obtain an image after anti-aliasing processing. And the anti-aliasing effect of the CT reconstructed image is improved.

The invention belongs to the technical field of building boundary segmentation, and provides a remote sensing image building boundary segmentation method based on a U-net network model, and the method comprises the steps: extracting a gradient magnitude image through a Sobel operator, generating a binary boundary mask, dynamically adjusting the weight of a convolution kernel in combination with a gradient direction, and enhancing the feature extraction capability of a boundary region; boundary artifacts output by cavity convolution are suppressed through a high-frequency residual boundary mask, an artifact area is restored in combination with gradient direction constraint, and original detail features are reserved; calculating complexity scores based on boundary density and tortuosity, adaptively selecting voidage, and balancing global context and local detail capture capability; extracting a high-frequency component through a Laplace operator, marking an artifact point by combining a dynamic threshold value, and repairing an abnormal region by utilizing a neighborhood gradient direction; the method solves the performance bottleneck of a traditional method in a boundary fuzzy, artifact interference and receptive field fixed scene, and is suitable for remote sensing image analysis tasks such as urban planning and disaster assessment.

The invention discloses an MFSE-UNet-based rail extraction method in unmanned aerial vehicle aerial images, relates to the technical field of unmanned aerial vehicle remote sensing image semantic segmentation, and solves the problems of fuzzy edges, poor multi-scale adaptability and weak generalization ability in complex scenes of rail extraction in unmanned aerial vehicle aerial images in the prior art. According to the method, a multi-scale feature fusion and low-resolution self-attention module is provided, multi-dimensional features are efficiently extracted, key information is enhanced, and segmentation definition and structural integrity are improved; a bidirectional feature fusion module is designed, detail and semantic information of an encoder, a decoder and adjacent layers are subjected to weighted fusion, semantic differences are bridged, and the multi-scale detection performance is improved; a learnable Sobel + EFR edge guiding mechanism is innovated, learnable characteristics of a Sobel operator and an EFR module are fused, complex scene edge characteristics are dynamically adapted, and track boundary positioning precision and segmentation definition are improved. According to the method, the rail extraction precision in a complex scene is effectively improved, and technical support is provided for intelligent detection and maintenance of railway tracks.

The invention discloses a method and a system for reconstructing a cavity shielded by a shallow reinforcing mesh. The method comprises the following steps: constructing a layered medium scene containing the shallow reinforcing mesh and the cavity; acquiring image data information with different shielding states and preprocessing the image data information to construct a reconstruction data set; constructing a reconstruction initial model of the cavity shielded by the shallow reinforcing mesh, and training to obtain a reconstruction model of the cavity shielded by the shallow reinforcing mesh; and adopting the obtained reconstruction model of the cavity shielded by the shallow reinforcing mesh to reconstruct the actual cavity shielded by the shallow reinforcing mesh. According to the method, multiple types of training data are acquired through a constructed layered medium scene containing a shallow reinforcing mesh and a cavity, and a reconstruction model of the cavity shielded by the shallow reinforcing mesh, which comprises a U-shaped network, a Sobel operator, an attention mechanism, a residual structure, an average mechanism and a jump connection scheme, is trained; therefore, the reconstruction of the cavity under the shielding of the shallow reinforcing mesh can be realized, the reliability is higher, the accuracy is better, and the reconstruction efficiency is higher.

The invention discloses an intelligent segmentation and feature analysis method for a medical image, and relates to the technical field of image analysis, and the method comprises the steps: reading a multi-modal medical image, and carrying out the preprocessing and size standardization; performing medical image segmentation based on a U-Net structure, performing edge detection by adopting a Sobel operator, and optimizing the edge of the segmented image in combination with morphological operation to obtain an edge-optimized focus segmentation map; feature point matching is carried out on the multi-modal focus segmentation image based on self-organizing mapping, and non-rigid registration is carried out through TPS transformation based on matched feature points; carrying out foreground region enhancement operation on the registered image, introducing a channel attention module based on a DenseNet-201 model to carry out feature extraction on the registered image, and optimizing features by using a whale optimization algorithm to generate an optimal feature subset; and carrying out illness state classification on the feature subsets by an adaptive neural fuzzy inference system optimized based on a genetic algorithm.

The invention provides an image multi-feature matching, tracking and positioning method, system and device for a rear axle assembly line and a medium, and belongs to the technical field of augmented reality processing of automobile rear axle assembly lines, and the method specifically comprises the steps: carrying out the image collection of a target at the automobile rear axle assembly line; calculating gradients of the image in the horizontal and vertical directions by using a Sobel operator to obtain gradient amplitudes and directions; matching the video frame with an offline collected template, and determining the initial position of the target in the video frame according to the similarity of the gradient descriptors; matching the ORB feature vector of the current image with an ORB feature vector of a pre-stored template image, and when the Hamming distance is smaller than a set threshold value, determining that the current image is a matched feature point pair; and the position of the target object is updated by minimizing the error of the feature points in the adjacent video frames. According to the method, the target can be accurately identified and tracked, and high-precision target position information is provided for quality detection, assembly guidance, process monitoring and other links in industrial automatic production.

The invention provides a MaskRcnn-based tumor detection method and system, and the method comprises the steps: inputting medical image data into a MaskRcnn network integrated with an edge perception module, generating an edge response graph through a Sobel operator, and enabling the edge response graph to serve as an additional channel injection feature graph, and enhancing the boundary representation; fusing multi-scale features through a feature pyramid network FPN, extracting candidate tumor area features through RoI Align, introducing non-local attention modeling global dependence in the area, extracting directional entropy, texture energy, uniformity and contrast in combination with a gray level co-occurrence matrix GLCM, and splicing to generate structure sensing features; performing classification, bounding box regression and mask segmentation based on the structure perception features, and constructing a joint loss function including classification, regression, mask cross entropy, edge alignment and structure consistency; the positioning capability of the model on the fuzzy tumor contour is effectively improved, the boundary positioning error is remarkably optimized, and the defect that over-segmentation or missing detection is likely to occur in a traditional model is overcome.

The application discloses a geotechnical engineering slope deformation monitoring method and system, through acquiring multi-modal image data, utilizing a CLAHE algorithm to enhance the contrast of weak deformation areas of the multi-modal image data, extracting gradient edge information through a Sobel operator, and filtering laser radar point cloud noise based on VMD variational mode decomposition; based on a BiLSTM bidirectional long short-term memory network, bidirectional dependence of a time sequence is captured, and weights of different time steps and spatial positions are dynamically allocated through an attention mechanism; the hyperparameters of the mixed prediction model are optimized by using a GJO golden jackal optimization algorithm to obtain a target mixed prediction model; feature image data is input into the target mixed prediction model for prediction to generate a risk probability heat map, and a potential sliding surface position of a slope is located according to the risk probability heat map. The reliability and accuracy of geotechnical engineering slope image data analysis are improved.

The invention provides an X-ray imaging detection method and device. The method comprises the following steps: correcting original image data of X-rays to obtain a corrected original image, extracting high-frequency information of the corrected original image through Gaussian blur, and performing difference and noise suppression processing to obtain preprocessed noise reduction data; carrying out binarization processing on the preprocessed noise reduction data to obtain noise reduction data; performing edge processing on the noise reduction data through a Sober operator to obtain an edge image; and after the edge image and the corrected original image are fused, the line is identified through Hough transform, a detection part image is obtained, the device comprises a module for realizing the method, through the method and the device, the ray machine device in the DR detection system is optimized, a high-frequency constant-voltage ray machine is replaced by a conventional power frequency ray machine, and the equipment cost is reduced.

The invention discloses a power transmission line icing thickness detection method and system based on deep learning, and relates to the technical field of image processing and power transmission line online detection, and the method comprises the steps: collecting an icing image of a power transmission line through a standardized dual-light optical device, and carrying out the image preprocessing, and generating an original data set; and performing image segmentation on the original data set, training the deep neural network model by using the lightweight YOLACT network, and performing evaluation. Edge detection is carried out through a multi-direction Sobel operator template, an Otsu method is adopted for binarization processing and nonlinear operator filtering denoising, a boundary chain code array is extracted based on a boundary tracking algorithm of a topological structure, and the MASK of an original power transmission line is determined. And acquiring an image of the power transmission line by using a dual-light camera, performing camera calibration, performing registration and fusion on the calibrated image, and outputting a calculation result of the icing thickness of the power transmission line by using the trained neural network. According to the invention, the detection precision and efficiency are improved, and the thickness estimation deviation caused by image errors is reduced.

This invention discloses a power transmission tower sign detection and recognition method based on CS-ABCNet, comprising the following steps: Step 1: Collect a dataset of power transmission tower signs; Step 2: Preprocess the dataset using the Sobel operator, calculating the edge information of objects in the image from both horizontal and vertical directions; Step 3: Extract features using ReXNet on the ABCNet backbone to achieve a lightweight network model. This method can preprocess tower sign features, optimize the network structure, use a lightweight convolutional neural network to obtain feature maps, and introduce a CBAM attention module into the detection head, thereby improving the recognition accuracy and performance of tower signs.

The invention discloses a natural gas hydrate CT image threshold segmentation method and a storage medium, and belongs to the technical field of natural gas hydrate detection.The method comprises the steps that S1, a natural gas hydrate sample image obtained through synchrotron radiation CT scanning is obtained, the image is preprocessed, and a central area image is obtained; s2, performing edge recognition on the preprocessed central region image by using a convolutional neural network and combining a Sobel operator to obtain an edge feature matrix; s3, clustering the edge feature matrix by adopting a KMeans clustering algorithm, and determining gray boundary values of the solid particles and the methane bubbles; s4, removing gray scale parts of solid particles and methane bubbles in the image according to the gray scale boundary value, and performing maximum and minimum normalization processing on the residual gray scale values of water and hydrates to obtain a gray scale probability density; and S5, drawing a normalized gray probability density map according to the gray probability densities of all the images, and determining a threshold segmentation range of the hydrate according to the probability density peak displacement. The method can automatically extract the hydrate threshold range, and improves the segmentation precision.

The invention discloses an endoscope video frame specular reflection highlight detection method and system, and belongs to the field of endoscope optical detection. The method comprises the following steps: firstly, converting an endoscope video frame into a grey-scale map; a Sobel operator is utilized to extract edge gradient to generate a binary image gradient mask, and meanwhile, a binary image highlight mask is generated by obtaining an area with the brightness obviously higher than the average level of a grey-scale map; dividing the binary image gradient mask and the binary image highlight mask into image blocks, adaptively adjusting the sizes of the blocks according to the highlight ratio of the image blocks so as to fuse the binary image gradient mask and the binary image highlight mask, and performing morphological processing and region screening to obtain a final binary image highlight mask of the grey-scale image; and carrying out time domain compensation on the final binary image highlight mask to obtain a specular reflection highlight detection result of the endoscope video frame. The accuracy, integrity and robustness of specular reflection highlight detection of the endoscope video frame are effectively improved.

The application discloses a reconstruction method and system for a cavity under a shallow reinforcement net shelter, and comprises the following steps: constructing a layered medium scene containing a shallow reinforcement net and a cavity; acquiring image data information with different shelter states and preprocessing to construct a reconstruction data set; constructing a reconstruction initial model for the cavity under the shallow reinforcement net shelter and training to obtain a reconstruction model for the cavity under the shallow reinforcement net shelter; and using the obtained reconstruction model for the cavity under the shallow reinforcement net shelter to perform actual reconstruction for the cavity under the shallow reinforcement net shelter. The application acquires multiple types of training data through the constructed layered medium scene containing the shallow reinforcement net and the cavity, and trains the reconstruction model for the cavity under the shallow reinforcement net shelter which comprises a U-shaped network, a Sobel operator, an attention mechanism, a residual structure, an average mechanism and a skip connection scheme. Therefore, the application can not only realize the reconstruction for the cavity under the shallow reinforcement net shelter, but also has higher reliability, better accuracy and higher reconstruction efficiency.

This invention provides a method and apparatus for detecting surface defects in steel based on an improved YOLOv11. The method includes: integrating an EA module into the backbone network of YOLOv11, combining the Sobel operator edge detection idea with an attention mechanism, significantly enhancing the network's ability to perceive the geometric edges of defects, effectively improving the detection performance of low-contrast defect-background boundaries, and enhancing the ability to identify key defects such as cracks; by integrating three MSA modules into the neck network, the model's ability to capture multi-scale information in the feature space is significantly improved; the special position design of the EA module enhances the network's attention to horizontal and vertical gradient features during the final fusion of multi-scale feature representations, avoiding excessive interference with original features or highly abstract features; and achieving end-to-end processing from the original image to the defect detection result, significantly improving the detection performance of low-contrast defects and cross-scale defects on the surface of steel equipment.

The invention provides an automatic component identification and process circulation method for a steel structure production line, and belongs to the technical field of steel structure production. The method comprises the following steps: firstly, constructing a feature fusion network, and inputting pre-collected images to obtain a feature set; after the feature set is visualized, convolution operation is carried out on the input signal and the convolution kernel to obtain a pre-training model; on the basis, a Sobel operator is adopted to carry out edge detection and image segmentation, then a defect range is constructed by taking a pixel point as a unit, and after the defect range is repositioned, a defect area is corrected according to a color value change trend. Finally, training is executed according to the data unit image, the number of the defect areas and the area range, a feature network model is obtained, and automatic recognition and real-time positioning of the defect workpiece are achieved. According to the invention, an unattended workpiece removing scheme is further designed, so that the continuity of process circulation is improved. According to the invention, more efficient quality monitoring is realized, the labor cost is reduced, and meanwhile, the overall production efficiency is improved.

The application discloses a ground cloud classification and identification method based on a multi-scale hollow convolution and a multi-channel attention mechanism, and provides a model as follows: a four-branch multi-scale texture feature extraction model is fused with a four-channel attention mechanism network model.In the four-branch multi-scale texture feature extraction model: the first branch performs linear combination on a channel dimension of a cloud picture pixel by pixel to capture point-level feature information; the second branch extracts local features such as cloud lines and flocculence; the third and fourth branches capture mesoscale features of cloud layer distribution and global cloud system spatial large-scale features respectively through hollow convolution.In the four-channel attention mechanism network model, a color channel extracts a channel brightness mean value through global average pooling; a texture channel measures pixel dispersion through global standard deviation; an edge channel uses a Sobel operator to detect a boundary line between clouds and the sky; and a contour channel uses a Laplace operator to capture contour details.The application significantly improves the accuracy of class judgment and the classification precision of complex cloud types.