14 results about "Sobel operator" patented technology

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.

This invention discloses an intelligent quality detection method and system for rainbow films based on image recognition, belonging to the field of image processing technology. The method includes convolving a rainbow film image using the Sobel operator to generate a binary mask, generating a filtered morpholuminescence map, performing a one-dimensional Fourier transform on the filtered morpholuminescence map, calculating the direction angle of the complex curl using the arctangent binary function, extracting the principal phase field, calculating the continuous phase field, convolving the continuous phase field to generate a smoothed continuous phase field, and calculating the gradient magnitude of the smoothed continuous phase field to generate a binary defect mask. This invention generates a high-quality morpholuminescence map through multi-level feature extraction and adaptive segmentation, enhancing the continuity of fringes and the distinguishability of defects. By constructing a smooth and continuous phase field and combining it with statistical thresholds for adaptive identification of defect regions, it improves the robust recognition capability for complex interference patterns.

The present application relates to the technical field of intelligent control and industrial visual detection of beneficiation process, and discloses an industrial-grade ore belt boundary identification method based on multi-frame time sequence fusion. A multi-branch hollow boundary enhancement unit is constructed based on deep convolution, different convolution kernel sizes, and a Sobel operator. A feature extraction and fusion module is constructed based on unbiased convolution, time sequence cross-channel difference, the multi-branch hollow boundary enhancement unit, and multi-head attention. A feature enhancement module is constructed based on deep separable convolution and the multi-branch hollow boundary enhancement unit. A prediction module is constructed based on the multi-branch hollow boundary enhancement unit. An identification model is constructed based on the feature extraction and fusion module, the feature enhancement module, and the prediction module. Continuous multi-frame pictures in an industrial field are acquired, and after preprocessing, the multi-frame pictures are input into the identification model to obtain ore belt boundary identification results. Based on the ore belt boundary identification results, sorting is performed, and the problem that existing table sorting and identification methods cannot effectively identify ore belt boundaries is solved.

The present application relates to radar data processing technical field, specifically to a kind of radar layer position tracking method of fusing confidence clustering and wavelet energy discrimination, comprising the following steps: S1, the gray processing is carried out to each frame radar image and the candidate layer position point of the edge gradient greater than threshold is extracted using Sobel operator;S2, detect and eliminate outlier, false edge and low confidence point;S3, according to confidence weighted clustering to generate structure point;S4, optimal trajectory is spliced using DTW algorithm to form layer position path;S5, calculate wavelet energy variation rate to correct deviated path;S6, curve fitting and smoothing to generate final layer position trajectory.The present application, by outlier elimination, confidence weighted clustering, dynamic path planning and wavelet energy auxiliary correction mechanism, improves the accuracy, continuity and robustness under complex structure region of radar layer position tracking.

The present application relates to a kind of printing texture graph topology analysis and vectorization method, the described method includes the following steps:1) the initial image is preprocessed;2) by Sobel operator and LBFGS++ algorithm, the multi-vector field model of image foreground point is established;3) the curve connected by foreground point is established, and according to multi-vector field model, curve tracking is carried out, and topological graph is formed;4) according to topological graph result, each group of curve is fitted by Bezier interpolation, and smoothing filtering processing is carried out, and finally output svg format vector file with line skeleton as graph element information.Compared with prior art, the present application has the advantages of high vectorization accuracy, wide processing range, etc.

According to a system for processing an unstructured multi-medical image registration algorithm and a method of using same, proposed in the present invention, the system comprises: a first image registration processing unit for performing image registration through a normalized cross-correlation (NCCS) process based on a Sobel operator; a second image registration processing unit for performing image registration through a normalized cross-correlation (NCCL) process based on a LOG operator; and a multi-resolution processing unit for performing image processing, with multiple resolutions, on each of registration images of the first image registration processing unit and the second image registration processing unit, thereby improving the accuracy of unstructured image registration by using a registration algorithm between unstructured multi-medical images using normalized cross-correlation based on the Sobel operator or normalized cross-correlation based on the LOG operator.

This invention discloses a method and apparatus for calibrating a telecentric structured light system based on the geometric center of a checkerboard pattern. The method includes: selecting test calibration board images from a set of calibration board images; performing image edge detection and binarization based on the Sobel operator to obtain a binarized edge map; performing line detection based on the Hough transform, calculating the rotation correction angles corresponding to the vertical lines of the checkerboard pattern, and constructing a rotation matrix using the rotation correction angles; constructing a standard checkerboard pattern template based on ROI selection and edge expansion processing of the test calibration board images; performing normalization processing on each calibration board image, and then performing template matching and sub-pixel-level center localization with the standard checkerboard pattern template to finally determine the center coordinates of the checkerboard pattern in the corresponding calibration board image as feature point coordinates; and using the rotation matrix to perform attitude correction on the feature point coordinates of each calibration board image, thereby calibrating the telecentric structured light system. This invention can significantly improve calibration accuracy and stability.

The application discloses a robot control method and device based on escalator detection, which comprises the following steps: collecting a to-be-detected image in real time by using a camera configured by a robot, processing the to-be-detected image by using a Sobel operator, a preset binary algorithm and image morphological operation, determining a plurality of gradient connected regions in the to-be-detected image, generating a minimum circumscribed rectangle frame by calculating whether the area of each gradient connected region is greater than a set threshold, obtaining position information of the minimum circumscribed rectangle frame, and controlling the robot to perform corresponding operation according to the position information of the minimum circumscribed rectangle frame. By adopting the embodiment of the application, the position of the escalator can be directly positioned, data processing is reduced, computer resources are avoided from being wasted, the accuracy of detection is improved through various processing of the to-be-detected image.

The application provides a layered image enhancement method based on a four-orientation neighborhood edge-preserving denoising operator, and belongs to the technical field of image processing. The method first realizes image adaptive layering by adaptive regularization parameter calculation and guided filter improvement based on a Sobel operator to obtain a base layer and a detail layer; secondly, a four-orientation neighborhood edge-preserving denoising operator is designed, a plurality of direction-sensitive 3*3 operators are fused into a single 5*5 composite convolution operator, and the detail layer is subjected to edge-preserving denoising treatment; finally, the base layer is subjected to contrast enhancement and fused with the processed detail layer to output an enhanced image. The application can effectively suppress noise while maintaining clear edges, and improve the overall quality and visual effect of an image.

This invention discloses a deep learning-based method and apparatus for segmenting the microstructures of high-temperature alloys, comprising: collecting images of the microstructures of high-temperature alloy materials from scanning electron microscopes; labeling and annotating the microstructures in the images; constructing a dataset through image preprocessing and image enhancement; constructing an improved Mask R-CNN model, employing adaptive histogram equalization to expand the dynamic range of image gray levels and enhance local contrast; fusing image edge information into the network model using the Sobel operator and introducing the Huber loss function to improve model performance; training the model through transfer learning; and using the trained instance segmentation model to identify and segment the SEM images to be analyzed, obtaining the segmentation mask and labeling rectangle for each microstructure in the image. This invention achieves accurate identification of microstructure objects in high-temperature alloy materials based on the improved Mask R-CNN model.

A method and apparatus for early warning of power equipment faults based on infrared image area expansion is disclosed. The method comprises the following steps: acquiring an infrared thermal image sequence of power equipment in a substation using an infrared camera; preprocessing the infrared thermal image sequence using morphological methods to solve for the difference image of the infrared thermal image sequence; performing edge detection on the two-dimensional matrix of the difference image using the Sobel operator to generate a new two-dimensional pixel matrix; scanning the new two-dimensional pixel matrix line by line to obtain a target region; extracting the endpoints and vertices of multiple contour curves from the target region; connecting the multiple contour curves based on the endpoint set and vertex set to obtain a closed curve; and solving for the area of ​​the closed curve; and realizing early warning or alarm of power equipment faults based on the change in the area of ​​the closed curve corresponding to the difference image. This invention reduces the probability of faults being obscured and thus advances the fault detection time.

The application discloses a tea leaf recognition method based on knowledge distillation and improved pruning technology, and comprises image acquisition and preprocessing, tea leaf images are acquired through an image acquisition device, color interference is eliminated through graying, horizontal and vertical gradients are calculated using a Sobel operator, low-quality images are screened through gradient amplitude variance, a threshold value M is set, images with a variance less than M are removed, input data quality is improved, a data set is constructed and enhanced, manual labeling is performed using a labeling tool, a data set is generated, the generated data set is given to an improved RT-DETR model for training, and a high-precision tea leaf picking teacher model is obtained; the application effectively compensates for the precision loss caused by pruning, makes the model perform excellently in the precision and speed trade-off, effectively compensates for the precision loss caused by pruning, performs excellently in the precision and speed trade-off, is superior to other pruning and distillation combination methods, and adopts mixed pruning of LAMP and channel pruning and multi-layer feature distillation based on a Mimic loss.

The present application relates to the tobacco leaf structure recognition technical field, especially to a kind of based on image recognition's tobacco sheet structure distribution's representation method, and include:1 through industrial camera collection loose after moisture tobacco light transmission image, and the light transmission image is carried out leaf segmentation processing, including: the original image is converted into gray image, and through binaryzation processing separates tobacco area and background, using Sobel operator extraction tobacco edge;2 in combination with gray scale segmentation algorithm, the tobacco gray scale image is folded layer of distinction, and according to edge information, the tobacco gray scale image area of different threshold is calculated;3 through the tobacco gray scale image area of different threshold and corresponding threshold, the structure distribution of tobacco after expansion is represented, provides reliable basis for the intelligent control of tobacco processing process.