167 results about "Superpixel segmentation" patented technology

The invention relates to the technical field of image recognition, and discloses a crop lodging image data detection method and system, and the method comprises the steps: carrying out the self-adaptive illumination correction processing of the original image data of crop lodging, and obtaining an illumination compensation image of the original image data; performing multi-scale super-pixel segmentation on the illumination compensation image to obtain a crop area image of the illumination compensation image; extracting texture features and shape features in the crop region image, and performing feature fusion on the texture features and the shape features to obtain a multi-modal fusion feature set of the crop region image; performing multi-dimensional feature collaborative analysis on the multi-modal fusion feature set to obtain an initial discrimination result of the multi-modal fusion feature set; performing confidence coefficient optimization on the initial judgment result in combination with a spatial context relationship to obtain a target lodging detection result of the crop region image; according to the invention, the efficiency of crop lodging image data detection can be improved.

The invention relates to the technical field of three-dimensional map recognition, and discloses a semantic segmentation-based low-altitude three-dimensional map element autonomous recognition method and system. The method comprises the following steps: acquiring a low-altitude remote sensing image and preprocessing to generate a multi-channel image matrix containing geographic coordinates and spectral characteristics; extracting multi-scale features through a pyramid feature extraction network in combination with cavity convolution, and obtaining an adaptive weighted feature tensor through a cascade attention mechanism fusion channel and a spatial weight; adopting a bidirectional feature fusion strategy to generate fusion features, and outputting an initial category probability distribution diagram by a semantic segmentation header network; obtaining a refined mask through edge perception optimization and superpixel segmentation correction, and mapping the refined mask to a three-dimensional coordinate system to generate a vector layer with a semantic tag; a constraint rule is deduced through a topological relation inference engine, logic conflicts are eliminated through rule-driven post-processing, finally, a standardized three-dimensional map element database meeting the geographic information standard is generated, and efficient, accurate and autonomous recognition of low-altitude three-dimensional map elements is achieved.

The invention discloses a non-reference image quality evaluation method and device based on multi-perception feature fusion and dynamic enhancement, and the method comprises the steps: obtaining the quality information of a multi-domain distorted image based on superpixel segmentation and Gaussian kernel filtering texture generation; quality related feature extraction is performed on multi-domain information through a semantic perception module and a distortion perception module, bidirectional modulation is performed on global visual features of an original distorted image through a cross attention mechanism, and dynamic fusion of multiple perception features is realized; through a parallel feature enhancement unit formed by local adaptive filtering of a visual self-attention block and a dynamic residual block, dynamic allocation of perception modes to different content areas is realized; generating a weighted quality score consistent with human visual perception through a weighted dual-path regression device; and outputting a predicted score consistent with the human score from the distorted image through the three sub-networks. According to the method, the problems of insufficient adaptability to complex content of a distorted image and low local distortion sensitivity are effectively solved.

The invention discloses a welding seam defect X-ray image intelligent diagnosis and credible traceability method based on multi-modal heterogeneous information collaboration, which is characterized in that a defect analysis network fusing multi-domain feature modeling and graph structure expression is constructed on the basis of bimodal data formed by a welding seam X-ray image and an industry detection standard text. In the image mode, dividing the weld seam image into a plurality of local area units through superpixel segmentation, taking the areas as image nodes, respectively extracting spatial domain, frequency domain, wavelet domain and edge domain features, and constructing a weighted graph structure by combining the spatial adjacency relation and the feature similarity relation between the areas; realizing overall modeling and correlation analysis of weld defect structure information by using a graph convolutional network; in a text mode, feature coding is carried out on an industry detection standard text, and the feature coding is used as an important prior constraint for defect judgment. Collaborative modeling of an image detection result and standard semantic information is achieved through a gating fusion mechanism, a mapping relation between a detection conclusion and a standard term is established, and interpretable expression and result credible traceability of the weld defect diagnosis process are achieved. And a welding seam X-ray film automatic digital acquisition and observation device is adopted in a matched manner, so that stable transmission, positioning observation and high-resolution digital imaging of the industrial ray film are realized, and reliable and consistent image data input is provided for the intelligent diagnosis method. The method is suitable for intelligent defect detection under complex welding seam structures and multi-working-condition imaging conditions, and has high engineering application value and popularization prospect.

The invention provides a milling cutter wear volume in-situ calculation method and system based on three-dimensional reconstruction, and relates to the technical field of machine tool cutter state detection, and the method comprises the steps: obtaining a high-quality multi-view image, employing an improved SLIC superpixel segmentation algorithm to extract region characterization parameters of a wear region superpixel block based on the high-quality multi-view image, and calculating the wear volume of a milling cutter according to the region characterization parameters. Inputting the region characterization parameters into a multi-scale fusion feature extraction network, and extracting a two-dimensional binarization matrix of the wear region; wear region feature points of the two-dimensional binarization matrix are extracted through a fusion feature detection algorithm, and a high-precision sparse point cloud of a wear region is generated through optimization by adopting a dual-stage feature matching algorithm; reconstructing the high-precision sparse point cloud by adopting an improved PMVS dense optimization algorithm to obtain a high-precision wear area point cloud; and based on the high-precision wear area point cloud, carrying out precise quantitative calculation on the wear area by utilizing a double-model fusion algorithm, and outputting to obtain a wear volume value. According to the invention, in-place accurate measurement and calculation of the wear volume of the milling cutter are realized.

The invention relates to the technical field of image data processing, in particular to a high-voltage switch shell surface coating uniformity evaluation method and system, and the method comprises the steps: collecting a shell surface coating image; dividing the image into a plurality of super-pixel areas by using a super-pixel segmentation algorithm introducing an adaptive distance measurement mechanism; extracting geometric structure features, texture features and adjacent color difference features of the target area; and carrying out nonlinear fusion on the adjacent chromatic aberration and geometric structure characteristics to construct inter-class characteristics, fusing the inter-class characteristics with texture characteristics serving as intra-class characteristics to obtain a comprehensive score, and judging whether the coating is uniform or not according to the comprehensive score. According to the method, the segmentation weight is adjusted in a self-adaptive manner, and the weak chromatic aberration is amplified in a nonlinear manner, so that accurate segmentation of a tiny defect region is realized; geometric irregularity, color mutation and texture anomaly features are integrated, detection of various tiny coating defects is achieved, and the evaluation accuracy is improved.

The invention relates to a laser radar point cloud densification method and device fusing image information, and a medium. The method comprises the following steps: fixing a visual camera and a laser radar on a rigid tool, and carrying out joint calibration and data alignment; a laser radar is used to collect three-dimensional point cloud data, and a visual camera is used to shoot a scene; performing super-pixel segmentation on the image by using an image brightness linear iterative clustering algorithm; mapping the three-dimensional point cloud data into a segmented two-dimensional image superpixel pattern spot region by using a jointly calibrated camera model conversion relationship to realize feature clustering of the original three-dimensional point cloud of the laser radar; performing curved surface fitting on a clustered result by using a random sampling consistency algorithm; and linear interpolation is carried out on the fitted curved surface, dense points which are not covered by the original point cloud of the laser radar are supplemented and generated, and densification of the point cloud of the laser radar is realized. According to the method, the point cloud densification precision and practicability are improved, and technical support is provided for automatic driving, robot navigation and three-dimensional reconstruction.

The invention relates to the technical field of remote sensing image processing, in particular to a multi-temporal remote sensing image change detection method and system based on a space-time diagram neural network, and the method comprises the steps: carrying out the geometric registration, radiation correction and superpixel segmentation of an image, and outputting a segmented superpixel region set; taking the superpixel of each time phase as a node, constructing a space-time diagram structure fusing the spatial adjacency relation and the multi-step time association, and generating a normalized space-time Laplacian matrix; spatial structure features and multi-temporal evolution features are extracted through a double-branch graph neural network, spatial branch and time branch output are fused based on a gating mechanism, and node embedding is generated; constructing positive and negative sample pairs based on node embedding, and optimizing a feature space by comparing a loss function; and calculating the Euclidean distance of node embedding at adjacent moments, and outputting a change detection result in combination with a dynamic threshold. The method effectively reduces the false alarm rate and omission rate, greatly improves the reasoning speed, and is suitable for the scenes of urban expansion monitoring, dynamic disaster evaluation and the like.

The invention belongs to the technical field of image processing, and particularly relates to a pet behavior analysis method based on image recognition, and the method comprises the steps: 1, obtaining a video frame sequence, carrying out the superpixel segmentation of each frame of image in the video frame sequence, and obtaining a superpixel image composed of a plurality of superpixel blocks; step 2, constructing a Poisson multi-Bernoulli hybrid tracking link by taking the region association graph as input, and segmenting a trajectory entity set into a time sequence fragment set; 3, constructing a skeleton * object * scene hypergraph by taking the time sequence fragment set as input; and 4, finally outputting a behavior analysis result carrying an object participation type and a scene type according to a feeding action segment and a water drinking action segment in the strong recognition initial result of the interactive behavior. According to the invention, unified modeling and interpretable behavior identification of pets, objects and scenes are realized, and the method has the advantages of high robustness, high precision and real-time performance.

The invention belongs to the technical field of cross-seasonal remote sensing image change detection, and particularly relates to a cross-seasonal remote sensing image domain adaptive change detection method based on an improved CycleGAN. According to the method, through segmentation of all models and boundary constraint multi-scale super-pixel segmentation, the source domain image and the target domain image are kept consistent in object-level structure, the problems of ground feature breakage, texture dislocation and sample alignment irregularity caused by seasonal differences are effectively reduced, pre-training ViT-B high-dimensional semantic features and a density clustering algorithm are introduced, and the accuracy and the robustness of the method are improved. Noise samples such as mixed ground features, shadows and illumination anomalies in a complex remote sensing scene can be automatically recognized, a source domain training set is made to be purer, the stability of CycleGAN style migration training is improved, a generator fuses a multi-scale residual block and a self-attention module, a migrated image is made to be close to a target domain in the aspects of color, texture and seasonal features, and the image migration efficiency is improved. And meanwhile, the boundary and the structure of the ground object are kept not to be damaged through semantic consistency constraint, so that the problem of false change in cross-seasonal change detection is fundamentally solved.

The invention provides a geographic information surveying and mapping data processing method and system, and the method comprises the steps: obtaining terrain superpixels, carrying out the segmentation operation of the terrain superpixels, obtaining terrain superpixel blocks, carrying out the pulse frequency coding operation of the terrain superpixel blocks, obtaining a pulse frequency sequence, and carrying out the pulse delay coding operation of the terrain superpixel blocks. The method comprises the steps of obtaining a pulse frequency sequence, obtaining a pulse delay sequence, adjusting pulse density through terrain superpixels, obtaining pulse density, carrying out feature integration on the pulse frequency sequence, the pulse delay sequence and the pulse density, and obtaining a structured feature tensor. The method comprises the following steps: obtaining original point cloud data, carrying out position calibration on the original point cloud data through equipment movement track data, carrying out feature extraction on original point cloud according to a structured feature tensor, obtaining a terrain block map with a classification label, and improving the accuracy of geographic information surveying and mapping data while ensuring terrain continuity.

The invention discloses a sandstone slice image component identification method based on a GL-SLIC algorithm and semi-supervised learning. The method comprises the following steps: firstly, acquiring a sandstone slice image amplified by 200 times under single polarization, converting an RGB image into a CIE-Lab color space, and extracting a GLBP texture feature vector; then a GL-SLIC superpixel segmentation algorithm is constructed, and adaptive segmentation based on texture and color features is realized in combination with a Gabor filter and a local binary pattern; then implementing a region merging algorithm to generate complete mineral particles and pore fragments; finally, a classifier based on VGG16 and a discriminator based on ResNet18 are constructed, a semi-supervised self-training framework is adopted, a model is initialized by using about 6% of manual labeling samples, a high-confidence-coefficient pseudo-label extension training data set is generated through iteration, and automatic recognition of sandstone components such as quartz, pores, kaolinite, rock debris and a matrix is achieved. According to the method, the recognition accuracy of 96.3% on a test set is achieved, compared with a traditional method, the segmentation precision and the recognition accuracy are remarkably improved, the data labeling cost is greatly reduced, and an efficient technical scheme is provided for automatic analysis of geological images.

The invention provides a fan blade background region adaptive segmentation method and system oriented to a complex environment. The fan blade background region adaptive segmentation method comprises the steps of extracting channels and edge feature operators of various color spaces; establishing illumination non-deformation transformation; detecting a sky area through a superpixel segmentation method, and performing adaptive adjustment; generating a blade probability graph by integrating multiple features; segmenting the blade probability graph to obtain a preliminary segmentation result; optimizing initial segmentation result energy by minimizing an energy function; and introducing a sub-pixel positioning method to perform edge refinement on the optimized preliminary segmentation result, and outputting a segmentation result. According to the invention, high-precision detection of the foreground and background areas of the fan blade is realized under the working conditions of a field open environment, various light environments, complex blade surface textures and the like; the problems that the background segmentation precision of an existing method in a complex field environment is insufficient, and the contradiction exists between the amplification factor and the small-size defect distinguishing capacity in fan blade non-stop detection are solved.

The embodiment of the invention discloses an automatic image segmentation method and device based on semantic segmentation and superpixel fusion, and the method comprises the steps: carrying out the image feature extraction of a target image, carrying out the semantic region division through a semantic segmentation module, and outputting a tongue region target image after region division; the method comprises the following steps: inputting a target image of a tongue region into a clustering engine for iterative calculation to obtain a super-pixel block set, setting a segmentation threshold value, performing pixel proportion calculation on the target image of the tongue region to determine a target affiliation category of the super-pixel block, and determining affiliation division of a tongue edge in the target image of the tongue region; according to the embodiment of the invention, semantic segmentation and pixel clustering are fused, an approximate tongue region is firstly positioned, and then the edge of the tongue is finely corrected, so that the edge blurring of pure semantic segmentation is avoided, and the semantic deviation of pure super-pixel segmentation is solved; meanwhile, a low segmentation threshold value is set to solve the under-segmentation problem of an edge fuzzy region, and the whole process is automatic without manual intervention.

The invention discloses a low-rank multi-modal remote sensing image clustering method and device on a superpixel manifold and a storage medium, and relates to the technical field of multi-modal remote sensing image clustering. The method comprises the following steps: splicing a multi-modal remote sensing image along a channel direction, segmenting the spliced image into a plurality of sub-regions through superpixel segmentation, and solving a mean value for each sub-region to obtain a multi-modal superpixel; embedding the Laplacian matrix of the multi-modal superpixels into manifold regularization about the superpixel clustering matrix, and capturing a local manifold structure of the multi-modal superpixels; under the constraint of manifold regularization, constructing a low-rank reconstruction model of a product of a single-mode clustering matrix and a unified clustering matrix; initializing and alternately optimizing the single-mode clustering matrix and the unified clustering matrix by using fuzzy clustering; and analyzing the super-pixel clustering result, and mapping the super-pixel clustering result into a clustering result of the original image. According to the invention, the accuracy and efficiency of remote sensing image clustering are improved.

The present application relates to the technical field of remote sensing image processing, in particular to a land consolidation boundary line division method and system based on machine vision. The present application obtains a plurality of superpixel blocks of a satellite remote sensing image; according to the elevation information distribution of different pixel points in each superpixel block in a digital elevation model, the terrain height variation degree of each superpixel block is obtained; according to the elevation information difference between each pixel point in each superpixel block and other pixel points in the corresponding neighborhood range in the digital elevation model, and the gradient distribution of the corresponding pixel points in different directions in the digital elevation model, the overall water flow direction value of each superpixel block is obtained; and then the merging possibility between different superpixel blocks is obtained; the superpixel optimal block is obtained, and the consolidation boundary line of the land region to be divided is divided. The present application accurately obtains the superpixel segmentation process of the land region to be divided, and improves the accuracy of the consolidation boundary line division.

The invention discloses a hyperspectral remote sensing image classification model based on double-flow feature collaborative modeling. The hyperspectral remote sensing image classification model comprises the following steps: step 1, carrying out superpixel segmentation; step 2, constructing a whole network architecture of the model; step 3, a detailed construction scheme of the model framework; step 4, a loss function fusion strategy; according to the hyperspectral remote sensing image classification model based on double-flow feature collaborative modeling, deep interaction and dynamic complementation of double-flow features in channel and space dimensions are achieved, global semantic information and local detail features are optimally integrated, and finally, through data set verification, the hyperspectral remote sensing image classification model based on double-flow feature collaborative modeling is obtained. The method is obviously superior to the existing mainstream method, shows superior small sample adaptive capacity and cross-scene generalization capacity, and provides effective technical support for intelligent interpretation of the hyperspectral remote sensing image.

The invention discloses a alumen ustum image clear area screening method based on superpixel segmentation and feature classification. The method comprises the following steps: generating a definition image of an original alumen ustum image by using a Laplace operator; enhancing the color contrast of the original alumen ustum image by using the definition image; segmenting the enhanced image into a plurality of small regions by using an SLIC superpixel segmentation method, and storing segmentation boundaries of all the small regions; applying the segmentation boundary to the original alumen ustum image, filling all small areas into a minimum enclosing rectangle, and storing the position and structure information of the minimum enclosing rectangle; using a ResNet feature extraction network to perform feature extraction on the small regions obtained by segmentation; and performing definition classification on all the small areas by using a linear classifier, and returning a classification result to an original image to obtain a clear alumen ustum image. According to the method, automatic identification and screening of clear alumen ustum structures in the image are realized through image small region division and region feature discrimination.

The invention provides a sea ice extraction method based on multi-feature superpixel segmentation, and the method comprises the steps: 1, obtaining a dual-polarized SAR image of a target sea area, executing the image preprocessing, and obtaining a preprocessed image; 2, backward heat dissipation coefficient features, standardized features, a morphological feature set, OTSU features and texture features are extracted from the preprocessed image and fused, and a multi-dimensional data feature set is obtained; step 3, segmenting the multi-dimensional data feature set by using an SNIC superpixel segmentation method to obtain a superpixel block set; and step 4, inputting the superpixel block set into a supervised and trained machine learning classifier, and classifying the seawater and the sea ice to extract the sea ice. According to the method, the sea ice classification precision and robustness can be effectively improved, a sea ice distribution feature set with the spatial resolution of 40 meters can be generated, and inter-annual and inter-monthly change monitoring of the north pole channel sea ice can be effectively supported.

The invention relates to the technical field of image processing, in particular to an unsupervised PolSAR classification method based on comparative learning, and the method comprises the steps: obtaining original PolSAR image data to form multi-view data; performing superpixel segmentation on the image to generate multi-view primitive samples in one-to-one correspondence with superpixels; obtaining potential feature representation of each view, and generating a self-representation coefficient matrix of each view through a forward self-representation module; applying cross-view relation consistency constraint by comparing loss, and optimizing a self-representation coefficient matrix of each view; fusing the self-representation coefficient matrixes of all the optimized views to generate a uniform affinity matrix, and executing spectral clustering to obtain clustering labels of element samples; according to the mapping relation between the element sample and the pixel established by the superpixel segmentation, the clustering label is mapped back to the pixel level, the image classification result is obtained, and the classification accuracy and robustness are remarkably improved.

The application discloses a magnetic resonance image segmentation method based on Dense-Unet and superpixels, and is characterized in that the method comprises the following steps: 1) data preprocessing; 2) improved UNet network; 3) superpixel segmentation; 4) result fusion; and 5) testing. The method has better semantic segmentation performance, can improve segmentation precision, and can segment more accurately in edge details.

This invention belongs to the field of computer vision and environmental monitoring technology, specifically relating to an environmental monitoring method and system based on images captured by unmanned aerial vehicles (UAVs). The method includes: acquiring images of a stockpile captured by the UAV and performing superpixel segmentation; extracting the color space statistical features, gray-level co-occurrence matrix features, and gradient features of each superpixel block; constructing a decoupling relationship for illumination saturation and using saturation differences to eliminate shadow interference; constructing a texture wetting smoothing relationship and using texture energy and gradient variance to eliminate interference from dark minerals; calculating an environmental compliance index based on the illumination saturation decoupling index and the texture wetting smoothing index, generating a heat map, and controlling the operation of dust suppression equipment accordingly. This invention, by decoupling physical optical features, effectively reduces misjudgments caused by backlighting, shadows, and the natural color of the ore, achieving accurate identification of the surface wetness of the stockpile, saving water resources, and improving dust suppression effectiveness.

The invention relates to the technical field of image processing, in particular to a plant raw material grading method and system based on image analysis. The method comprises the steps of collecting an RGB image of a to-be-detected frozen plant raw material, converting the RGB image into a CIELAB color space to obtain a corresponding color texture image, and performing super-pixel segmentation on the color texture image to obtain a plurality of super-pixel blocks; based on the brightness feature, the gradient feature and the saturation feature of each super-pixel block, obtaining a frost shielding index of each super-pixel block, and obtaining a final quality score of each super-pixel block according to the saturation feature of each super-pixel block and the difference between the gradient feature and the frost shielding index; taking the superpixel blocks belonging to the same berry as each superpixel block set; the areas of the super-pixel blocks serve as weights, the weighted quality score of each super-pixel block set is obtained, grading processing is conducted on each fruit based on the weighted quality scores, and the grading accuracy of frozen plant raw materials is remarkably improved.

The invention discloses an image segmentation method based on superpixel feature optimization graph segmentation. Comprising the following steps: firstly, acquiring and obtaining an original scar image through a camera, and then preprocessing the original scar image to obtain a preprocessed image; carrying out superpixel segmentation on the preprocessed image and extracting mean feature vectors corresponding to different superpixel areas; performing mask processing on the mean value feature vectors corresponding to all the superpixel regions to obtain rough segmentation masks; and finally, taking the rough segmentation mask as an initialization seed, and performing pixel-level edge refining on the original scar image by using an image segmentation method to obtain a refined segmentation mask. The method has the advantages of high segmentation precision, strong anti-noise capability, no need of manual seed marking and high automation degree, and can meet the actual application requirements of clinical diagnosis and quantitative analysis.

The present application belongs to the technical field of space-to-earth observation (satellite remote sensing) and watershed hydrology, geographical environment disciplines, and discloses a watershed scale high spatio-temporal resolution river network remote sensing information extraction method and system. After cloud and cloud shadow processing of optical remote sensing images, a simple non-iterative clustering superpixel segmentation algorithm is used to segment homogeneous landscape objects, and a hierarchical decision tree is constructed using a water body index and a vegetation index. In the area where the optical remote sensing image is seriously covered by clouds or the image is missing, the SAR image is filtered, the dual polarization water body index is calculated using the VV and VH polarization images, the maximum inter-class variance method is executed, and the image is binarized. After obtaining the water body result extracted from the remote sensing image, the river cut-off is repaired in combination with the water system diagram generated by the DEM model. The present application can automatically extract accurate, continuous and high spatio-temporal resolution river network remote sensing information, and can be expanded to global scale river network feature extraction.

The invention relates to the field of image data processing, in particular to a ring-scan sonar two-dimensional imaging data processing method and system, and the method comprises the steps: obtaining a ring-scan sonar image, carrying out the superpixel segmentation of the ring-scan sonar image, taking any superpixel block as a target block, calculating the average noise pollution degree and the average echo signal intensity of the target block, and obtaining the ring-scan sonar two-dimensional imaging data. Taking the ratio of the average echo signal intensity to the average noise pollution degree as the comprehensive weight of a target block, taking the comprehensive weight of the target block as the comprehensive weight of each pixel point in the target block, and traversing to obtain the comprehensive weight of each pixel point of each super-pixel block, constructing a weight map aligned with the ring scan sonar image according to the comprehensive weights of all the pixel points of all the superpixel blocks; and taking the ring scan sonar image and the weight map as dual-channel input of a preset detection model to obtain a detection result. According to the invention, the target detection precision of the ring scan sonar is improved.

The invention belongs to the technical field of marine geological exploration, and particularly relates to a seabed sediment classification method and system based on a side-scan sonar image, and the method comprises the steps: introducing an acoustic imaging physical mechanism based on a Lambertian reflection model, carrying out the inversion of acoustic parameters of seabed sediments, and revealing the cause of echo intensity from the physical level; meanwhile, a superpixel segmentation algorithm is combined, analysis units with consistent structures are formed in space, noise interference is effectively reduced, and the salt and pepper phenomenon in pixel-level classification is avoided; furthermore, multi-source fusion is performed on image features, reflection features and topographic features, and the precision, the spatial continuity and the physical interpretability of seabed sediment classification are remarkably improved by utilizing the complementary advantage of multi-dimensional information. The method is suitable for seabed sediment classification in a complex terrain environment. The invention also provides a non-transitory readable recording medium storing the program of the method and a system comprising the medium, and the program can be called through a processing circuit to execute the method.

Embodiments of the present application provide a kind of image restoration method, device, electronic equipment, storage medium and program product.The method comprises: obtaining to-be-processed image, and based on superpixel segmentation technology, to-be-processed image is carried out multi-granularity segmentation processing, obtain multiple segmented to-be-processed images;According to the multiple image blocks included in each segmented to-be-processed image, to-be-processed image is carried out noise identification processing, obtain the result of noise identification processing, according to the result of noise identification processing and to-be-processed image, generate multiple noise mask images;Wherein, each noise mask image indicates different granularity;The granularity indicated by noise mask image corresponds to the granularity indicated by segmented to-be-processed image;According to multiple noise mask images, to-be-processed image is carried out noise restoration processing, obtain the to-be-processed image of repair completion.The method is used to reach the effect of improving the accuracy of image restoration.

The invention discloses a mite removal area positioning method based on image recognition, and the method comprises the following steps: S1, collecting RGB, infrared and depth images, extracting surface features, and generating a high-dimensional feature tensor field through manifold learning; s2, performing superpixel segmentation on the tensor field to obtain an analysis region and calculate an attention weight; s3, generating a multi-scale deformable tensor kernel according to the structural features and the neighborhood relationship; s4, inputting the analysis region, the tensor kernel and the neighborhood relationship into a voting field of the spatial domain and the frequency domain to generate a fusion voting response; s5, training a tensor kernel parameter and a voting strategy based on the fused voting response to obtain an optimized parameter; s6, combining the optimization parameters, fusing the voting response and the attention weight, and calculating an abnormal probability; and S7, carrying out weighted fusion on the abnormal probability and the fusion response, and outputting an acarus killing area positioning result. According to the method, the multi-mode features of the image and a tensor mechanism are fused, and accurate positioning and structure self-adaptive modeling of the acarus killing area are achieved.

The application discloses an RGB-D saliency detection method and device, electronic equipment and medium, wherein the method comprises: preprocessing an input image; generating a superpixel segmentation map and extracting a boundary; inputting the preprocessed RGB image and depth image into an encoder to obtain multimodal features; fusing the RGB features and depth features; inputting the fused features into a decoder for decoding to output a predicted saliency map. The RGB features are input into a boundary perception module to output a predicted RGB image boundary map; the fused features are input into the boundary perception module to output a predicted saliency map boundary; the saliency target prediction result is supervised by using a fine-labeled segmentation map, the RGB image boundary is predicted by using the boundary map, and the learning of the model is guided by using a loss function. The application provides boundary guidance for the network through a superpixel generation algorithm, greatly improves the perception of the network to the edge, effectively improves the segmentation quality, and can be widely applied to the field of image processing.