589 results about "Edge enhancement" patented technology

The invention relates to the technical field of photoetching machine calibration, and discloses a photoetching machine calibration method, device and equipment based on multi-view vision, and the method comprises the following steps: carrying out imaging and phase sensitive detection analysis on a mask plane and a wafer plane through a four-path optical beam splitting system to obtain four groups of mask-wafer initial alignment position information; performing scanning white light interference edge enhancement processing to obtain three-dimensional surface contour data; performing wavelet transform processing on the three-dimensional surface contour data, and performing dynamic registration on the four-path optical beam splitting system to obtain multi-view visual feature registration data; a comprehensive error model including mechanical errors, optical errors and environmental errors is established, real-time correction and error compensation are carried out on a six-degree-of-freedom motion platform of the photoetching machine, a multi-view vision calibration result of the photoetching machine is obtained, the influence of environmental vibration and thermal drift on calibration precision is effectively eliminated, and the calibration accuracy of the photoetching machine is improved. The optical system is ensured to be always in the optimal imaging state, and the calibration precision is improved.

The invention provides a liver focus three-dimensional modeling method, and belongs to the technical field of image processing based on computer vision. Firstly, a multi-view spatial registration method based on optical flow optimization is designed, pixel-level displacement information of different view images is estimated by calculating an optical flow field, accurate image alignment is achieved, and spatial consistency of three-dimensional reconstruction is improved. And secondly, a three-dimensional reconstruction strategy based on two-dimensional focus segmentation is proposed, the two-dimensional focus segmentation is completed by adopting a lightweight U-Net variant, and a segmentation result is mapped to a three-dimensional space through a voxel probability projection method, so that 3D focus reconstruction is realized, and the calculation cost is reduced. And finally, extracting high-frequency features of the three-dimensional model by adopting a local edge enhancement method based on a Laplacian operator, and strengthening a focus boundary and a key anatomical structure through interpolation optimization, so that the three-dimensional model is more accurate and clearer. Compared with a traditional method, the method has the advantages that the mode of purely depending on image superposition is avoided, and the accuracy of three-dimensional modeling is improved.

The invention relates to the technical field of image processing, and provides an enteromorpha remote sensing image detection method and system based on deep learning, and the method comprises the following steps: carrying out the edge gradient extraction and small target enhancement processing of an obtained to-be-detected remote sensing image, and obtaining a first feature map after the edge enhancement and small target feature enhancement; transmitting the first feature map to a U-shaped backbone network formed by cascading multiple stages of Ep-VSS block modules, performing multi-stage feature extraction, and obtaining a detection result of the enteromorpha remote sensing image based on the feature map output by the last stage of Ep-VSS block module; according to the method, through edge enhancement, small target sensitive detection, multi-scale texture extraction and spatial context modeling, precise boundary segmentation and long-range dependence modeling are realized, and the accuracy, real-time performance and reliability of enteromorpha remote sensing monitoring are improved.

The invention discloses a remote sensing image fishpond extraction method based on an adaptive edge enhanced neural network. The method comprises the following steps: selecting a sentinel No.2 satellite multispectral band to synthesize a false color image; constructing a fishpond region labeling data set; establishing an adaptive edge enhanced neural network architecture, wherein the architecture fuses an edge detection module capable of learning a threshold and semantic feature extraction; an edge perception double attention module and a pyramid pooling module are integrated, and the multi-scale feature representation capability is enhanced; optimizing a training process by adopting a multi-level depth supervision and prediction fusion strategy; and finally, extracting a fishpond area through the training model and generating a standardized result. According to the method, the fishpond boundary segmentation precision is remarkably improved, the adaptability to fishponds with different scales is enhanced, the detection rate of small-scale fishponds is particularly improved, the model convergence process is accelerated through the depth supervision strategy, misclassification of similar water bodies such as rivers and ditches is effectively reduced, and the accuracy of fishpond boundary segmentation is improved. The technical problems that in the prior art, edges are fuzzy, multi-scale adaptability is poor, complex background interference is sensitive, and the omission ratio of a small-scale fishpond is high are solved.

The invention provides a rock image classification method based on edge enhancement and multi-scale feature fusion. By combining the edge enhancement and multi-scale feature fusion technology, the accuracy of mineral classification in the rock slice image is remarkably improved. According to the method, rock slice image interference is eliminated through filtering and denoising, and mineral particle edges are enhanced by fusing morphological top-hat transformation and a Laplace operator; and performing multi-scale pyramid decomposition on the enhanced image to extract high-frequency information, performing multi-direction response enhancement to generate a direction feature map, and performing channel-level fusion on the original image, the edge enhanced image and the direction feature map. And finally, inputting the fused image into a convolutional neural network to complete rock type classification. According to the method, mineral boundary expression is effectively enhanced, the classification accuracy is improved, and a reliable technical scheme is provided for geological analysis and lithology identification.

The invention provides a fuzzy barcode image processing method and system fused with super-resolution repair. The method comprises the following steps: acquiring a bar code image containing dynamic blurring, detecting a blurred region formed by superposition of motion tracks in the bar code image based on space-time continuity characteristics of the dynamic blurring, extracting an edge diffusion direction of the blurred region, and determining the dynamic blurring of the bar code image according to the edge diffusion direction. And performing multi-scale detail layer generation on the fuzzy region, inputting the multi-scale detail layer into a fusion repair module, and generating a repaired high-resolution barcode image by alternately executing super-resolution reconstruction guided by local high-frequency information and morphological repair of a module gap. Edge sharpening and noise suppression are carried out on the high-resolution barcode image, and a clear image meeting the barcode decoding standard is output; according to the technical scheme provided by the invention, the restoration precision and reliability of the complex dynamic fuzzy scene are remarkably improved, and the output image strictly meets the geometric constraint requirement of bar code decoding.

The invention discloses an efficient target detection method in a foggy environment, and relates to the technical field of target detection. According to the method, a lightweight foggy day target detection model is established and called DF-DETR, a defogging module of a double-branch structure is designed through an edge enhancement module, and target features of a foggy day image are effectively captured; then, a dual convolution feature extraction module DualConv-Block is designed, so that feature extraction is enhanced, and meanwhile, the complexity and the calculation amount of the model are remarkably reduced; besides, an EAA attention mechanism is combined with an intra-scale feature interaction module to form an AIFI-EAA module, and the AIFI-EAA module is integrated into the hybrid encoder, so that the attention capability of the model on dense targets is improved, and missing detection and false detection are effectively reduced; finally, a dynamic sampling scale attention feature fusion module is designed, alignment of multi-scale features is achieved through dynamic up-sampling, the flexibility and robustness of feature expression are enhanced, and the fusion and expression ability of the multi-scale features is further optimized.

The invention relates to the technical field of image enhancement, and discloses an image enhancement method based on an ISP image signal processing visual sensor, and the method comprises the steps: carrying out the collection of multiple frames of images of the same scene, carrying out the analog-to-digital conversion of the collected multiple frames of images, generating an RGB three-channel digital matrix, and obtaining N frames of original image sequences; dark current correction and lens shadow correction are executed, an affine transformation matrix from the N-1 frame to the first frame of image is calculated based on SIFT feature point extraction and matching, and an image sequence after space alignment is obtained; performing spatial feature extraction to obtain a multi-scale feature map sequence, and performing residual feature connection operation to obtain a time sequence correlation feature map sequence; according to the method, attention calculation of the channel dimension and the space dimension is executed, meanwhile, structural layer edge enhancement and texture layer denoising processing are executed, a target enhanced image is obtained, various kinds of distortion and noise interference in the image acquisition process are reduced, and the definition and the detail fidelity of the enhanced image are ensured.

The invention discloses an industrial image change anomaly detection method and system based on artificial intelligence, and relates to the technical field of image recognition, and the method comprises the steps: collecting a dual-light-source industrial image, employing frequency domain saliency to guide fusion, and carrying out visual enhancement processing through color mapping and edge enhancement; inputting the enhanced image into a CNN convolutional network to generate a multi-scale feature map, extracting a multi-scale high-pass residual image through two-dimensional fast Fourier transform and a high-pass filtering template, and splicing all scales and coding to generate a token sequence through local attention guide fusion; constructing a self-induction visual model, and performing feature reconstruction on the token sequence to generate a reconstructed feature map; and calculating and reconstructing an error scoring graph by adopting double error indexes, sampling to obtain an abnormal smooth graph, and segmenting an abnormal region based on the abnormal smooth graph. And finally, a dual anomaly detection system of image-level judgment and region-level identification is constructed.

The invention discloses a carton printing defect detection method and system based on visual identification, and relates to the field of defect detection, and the method comprises the steps: obtaining a printing carton image set; constructing an edge enhancement detection model based on YOLov; taking the image set as input, and extracting a plurality of feature maps with different resolutions by using a backbone network; inputting a plurality of feature maps with different resolutions into a neck network, carrying out adaptive aggregation through a feature pyramid module to obtain an aggregated feature map, then obtaining a direction and an edge response weight through an edge sensing module, and carrying out edge reservation fusion to obtain an edge enhancement feature; and inputting the edge enhancement features into a detection network, and obtaining defect detection positions and types through multi-task training of position perception loss and direction enhancement classification loss. Aiming at the low detection precision of the printing defects of the carton, the detection precision of the ink dot and broken line defects is improved through an edge reservation fusion strategy, adaptive feature fusion and the like.

The invention relates to the cross technical field of computer vision and civil engineering, and discloses a deep learning-based bridge disease image segmentation method, which comprises the following steps of: establishing an image data set containing crack and spalling diseases and performing online enhancement; constructing a segmentation network model comprising a frequency dynamic convolution encoder branch, an edge enhancement Transform encoder branch, a gating cooperation unit, a decoder and a depth supervision module; training the model by using a weighted mixed loss function; and inputting the test set to obtain a final segmentation mask. Self-adaptive fusion of local texture features and global context information is realized through a dual-encoder architecture and a gating cooperation mechanism; a frequency dynamic convolution and edge enhancement module is utilized to enhance the anti-noise capability and micro-disease perception under a complex background; and in combination with a category weighting strategy, the problem of pixel category imbalance is effectively solved, and high-precision automatic segmentation of concrete bridge diseases is realized.

The invention discloses a seal removing and document repairing method based on quantum state collaborative regulation and control, and relates to the field of document image processing and quantum computing cross technology, the method comprises the following steps: obtaining to-be-processed information, and carrying out quantum-classical feature collaborative preparation; the character stroke continuity is guaranteed through quantum entangled state modeling and entanglement degree constraint iteration, texture decoupling is achieved through quantum wavelet transform and Gram-Schmidt orthogonalization, and adaptive filling is conducted in combination with a quantum generative adversarial network; dynamic quantum phase adjustment is used for counteracting superposition interference of stamps with different transparency, and quantum neural network noise reduction and multi-scale quantum Fourier sharpening are used for optimizing image quality; checking the repair result, if the repair result does not reach the standard, returning to the edge sharpening link to perform decoupling and filling the edge sharpening link to readjust the parameter; and for special scenes such as inclination, multi-color overprinting and ultra-thin frames, quantum rotation correction, color channel separation and boundary annihilation operator processing are used, finally, high-precision, high-naturalness and high-adaptability restoration of seal removal is achieved, and high fidelity of results is guaranteed.

The invention provides a high-reflection plane metal part defect detection method and system based on a composite polarized light source, and a medium. The method comprises the following steps: illuminating a high-reflection plane metal part based on the composite polarized light source, adjusting a light source polarization angle and an LED wave band combination to construct a shooting environment, and collecting an image sequence; calculating a polarization intensity distribution parameter for the image sequence based on the relationship between the polarization intensity and the Stokes parameter to obtain a polarization intensity distribution image; analyzing the polarization intensity distribution image based on a polarized light reflection model, and outputting a specular reflection component and a diffuse reflection component; fitting the specular reflection component and separating the diffuse reflection component to obtain a reflection-removed image; based on an image enhancement algorithm, edge enhancement is carried out on the reflection-removed image to obtain an enhanced image, and based on a defect identification algorithm, the enhanced image is analyzed to obtain defect information; by removing the diffuse reflection component and reconstructing the image, the mirror interference is effectively stripped, and the defect identification precision is improved.

The invention belongs to the technical field of image processing, relates to a multi-scale road crack detection method and device in extreme weather and a medium, and relates to the technical field of image processing, and the method comprises the steps: constructing an extreme weather data set; constructing a target detection network; training a target detection network by adopting the data set to obtain a target recognition model; according to the method, the lightweight backbone network CAPNet is adopted to replace a backbone network of the RT-DETR, and the feature extraction efficiency is improved; a cascade multi-branch feature fusion module is used for replacing a CCFM module in a traditional neck network, dynamic integration of multi-scale features is achieved, a multi-scale edge enhancement module is additionally arranged in a network structure, the perception ability of the model to crack edges of different scales is enhanced, the feature expression problem under the low contrast and blurred image conditions is effectively solved, and the dynamic integration of the multi-scale features is achieved. And the balance between the detection precision and the efficiency is realized.

The invention discloses a photovoltaic module infrared image fault detection method based on unmanned aerial vehicle inspection, which relates to the field of fault detection, and comprises the following steps: configuring an unmanned aerial vehicle platform, planning a flight path, setting aerial photography parameters, setting an infrared thermal image acquisition and infrared image data return and storage mechanism, and performing gray normalization processing, image noise reduction, histogram equalization, edge enhancement processing and size standardization processing on the infrared image data, and performing data enhancement operation. An unmanned aerial vehicle infrared inspection technology is combined with a deep learning target detection model, a set of complete photovoltaic module infrared image fault detection process is established, and full-process automatic processing from image acquisition, image preprocessing, model detection to result evaluation and visualization can be realized. The method has the comprehensive advantages of being high in fault recognition precision, high in detection speed, standardized in processing flow and the like, and the efficiency and the intelligent level of photovoltaic power station component-level fault inspection are remarkably improved.

The invention relates to the technical field of medical image processing, in particular to a cross-modal image fusion detection system for endoscopic early cancer lesions, which comprises a data acquisition module for acquiring a white light endoscope and a narrow-band imaging image; the data annotation module carries out pixel-level annotation, and the data enhancement module amplifies a data set by using CycleGAN; the white light data module and the narrowband data module train feature extraction models respectively; the feature alignment and attention module solves the problem that feature scales and semantics between two modes are inconsistent, and efficient fusion is achieved. The edge enhancement module enhances feature extraction of small focuses through the generative adversarial network; the fusion model construction module fuses the features and the transition area data to generate a fusion model; and the output module processes the white light and the narrow-band imaging image and outputs a fused image. Through feature alignment, an attention mechanism and a CyclGAN data enhancement technology, the problem of insufficient medical image data is effectively solved, the model generalization ability is enhanced, and efficient and accurate early cancer focus detection is realized.

The invention discloses a multi-scale SAR (Synthetic Aperture Radar) image ship detection method and system based on edge enhancement and diffusion denoising, and mainly solves the problems that the existing SAR ship detection method is sensitive to noise and poor in small target feature extraction capability. According to the implementation scheme, the method comprises the following steps: obtaining an SAR image, carrying out standardized preprocessing, inputting the SAR image into a deep convolutional neural network, extracting a multi-scale feature map, and carrying out dynamic channel fusion enhancement on a low-layer feature map in the multi-scale feature map to obtain a fused high-quality feature map; performing differential edge enhancement on middle and high-level feature maps in the multi-scale feature map, and inputting the enhanced feature map and the fused feature map into a diffusion model detection head for training; and inputting a to-be-detected SAR image into the trained diffusion model detection head, outputting a preliminary target bounding box and a category confidence coefficient, performing score screening and non-maximum suppression operation on the preliminary target bounding box and the category confidence coefficient, and generating a final ship target detection result. According to the method, the precision and robustness of SAR image ship detection are remarkably improved, and the method can be used for ocean monitoring and military reconnaissance.

The invention discloses a multi-scale frequency-space fusion camouflage target detection method and system, and belongs to the technical field of camouflage target detection, an image and wavelet transform of the image are input into a WaveCamoNet model for camouflage target detection, and the model comprises a double-flow feature extraction module, a cross-domain dynamic fusion module, an edge texture enhancement module and a hierarchical decoder; inputting the image and the wavelet transform of the image into a double-flow feature extraction module, and extracting multi-scale spatial domain features and frequency domain features; inputting the multi-scale spatial domain features and the frequency domain features into a cross-domain dynamic fusion module, and performing multi-scale fusion and cross-domain dynamic fusion to obtain multi-scale fusion features and cross-domain dynamic fusion features; inputting the cross-domain dynamic fusion features into an edge texture enhancement module, and performing spatial calibration to obtain edge enhancement features; and inputting the multi-scale fusion features, the cross-domain dynamic fusion features and the edge enhancement features into a hierarchical decoder for decoding to obtain a binary detection image, and completing pixel-level positioning of the camouflage target.

The invention discloses an infrared thermal imaging gas leakage image identification and positioning method based on deep learning. The method comprises the following steps: S1, capturing thermal signal characteristics of gas leakage through an infrared thermal imaging module; s2, performing bilateral filtering and edge enhancement processing on the infrared thermal imaging image through a preprocessing module; s3, a deep learning detection module is improved through a YOLOv8-SAM2 model; and S4, the optical flow tracking and positioning module calculates a continuous frame optical flow field based on a Lucas-Kanade algorithm, reversely deduces the coordinates of a leakage source, and realizes three-dimensional space positioning in combination with GPS / IMU (Global Positioning System / Inertial Measurement Unit) data. The infrared thermal imaging gas leakage image recognition and positioning method based on deep learning solves the problems that in infrared thermal imaging gas leakage detection, tiny leakage recognition under a low-contrast image is difficult, the false alarm rate under complex background interference is high, and the real-time positioning precision is insufficient.

The invention relates to the field of image processing, and particularly discloses a rural highway pavement disease intelligent identification and positioning system comprising an image standardization module used for obtaining a standardized grayscale image; the wavelet decomposition module is used for obtaining a low-frequency sub-band and a plurality of high-frequency sub-bands; the high-frequency processing module is used for carrying out soft threshold processing on the high-frequency coefficient and retaining high-variance region features; the inverse transformation module is used for reconstructing the de-noised image; the edge enhancement module is used for highlighting crack and pit slot target contour information; the binarization module is used for segmenting a foreground disease candidate region; and the edge filling module is used for separating the disease from the background. According to the method, the core contradiction between background impurity removal and disease feature retention in rural highway tiny disease recognition is effectively solved, a traditional denoising algorithm either smooths tiny disease features or cannot thoroughly remove background impurities, and the system achieves the balance of the background impurity removal and the disease feature retention through cooperation of multiple modules.

The invention relates to the technical field of vision assistance, and discloses an anesthesia puncture positioning method and system based on vision assistance, and the method comprises the steps: accurately positioning an anesthesia puncture point through multi-view image fusion, hyperspectral image processing, image preprocessing, illumination equalization, edge enhancement, depth feature extraction and the like. The method comprises the following steps: firstly, constructing an image coordinate system, collecting a plurality of camera images, and obtaining a fused clear image through a designed multi-view fusion algorithm and distance calculation; and in combination with a hyperspectral image fusion algorithm, the image quality is further improved. Then, residual mapping filtering denoising and local histogram enhancement are used for illumination equalization, and the image contrast and edge details are enhanced; a convolutional neural network is adopted, interest point detection is carried out, a Hessian matrix is utilized to describe image second-order changes, and local depth features are extracted. And accurate positioning of an anesthesia puncture point is realized through a weighted soft voting classifier and a dynamic threshold method.

The invention relates to the field of steel surface detection, in particular to a steel surface defect detection method based on multi-scale edge enhancement, steel surface defect detection is performed through a constructed defect detection model, and the defect detection model is established through the following steps: constructing an MEF-NET model which is an improved YOLOv 11 network model, the MEF-NET model comprises a backbone network for carrying out feature extraction on an input image, a neck network for carrying out multi-scale feature extraction and fusion on a feature map and a head network for carrying out task specific output on fusion features provided by the neck network, a C3k2 module of the original backbone network is replaced by an AHED module, an SPPF module is replaced by an MD-FPN module, and the head network is used for carrying out task specific output on the fusion features provided by the neck network. A C2PSA module is replaced by a C2BRT module, a detection head of an original YOLOv11 network model is replaced by Ehead, and the MEF-NET model provided by the invention obviously improves the detection performance.

The invention discloses an unmanned aerial vehicle small target detection method based on feature enhancement and selective attention and a related device, and relates to the technical field of computer image target detection, and the method comprises the steps: obtaining an aerial image of an unmanned aerial vehicle, adjusting the image to a preset resolution, and obtaining an adjusted image; inputting the image into a backbone network of a network architecture, extracting a multi-level feature representation through a plurality of CALBlock feature extraction modules, and enhancing features of the multi-level feature representation through an EMIT edge enhancement architecture to obtain an enhanced multi-level feature representation; inputting the enhanced multi-level feature representation into a neck network of the network architecture, and performing cross-scale feature fusion and enhancement through an FSAFPN structure to obtain a multi-scale enhanced feature map; and inputting the multi-scale enhanced feature maps into a detection network of a network architecture, and performing target detection by using a decoder to obtain target category probability distribution and bounding box coordinates.

The invention relates to the technical field of wall thickness detection, in particular to a capsule wall thickness online detection system and method, and the system comprises an illumination correction module, an edge enhancement module, a wall thickness calculation module, an abnormal trend analysis module and an early warning feedback module. According to the invention, by analyzing and adjusting the non-uniform illumination area, the edge features of the capsule are clearer, the imaging error is reduced, the pixel points with the gradient change rate meeting the set threshold value are screened as the wall thickness boundary points, the wall thickness data deviating from the set threshold value are marked and removed, the stability of wall thickness detection is improved, and the wall thickness numerical value sequence of batches of capsules is obtained. Calculating the fluctuation degree of the wall thickness, analyzing the fluctuation trend of the wall thickness between batches, marking the batches with abnormal wall thickness fluctuation, providing trend early warning capability, combining the fluctuation trend of the wall thickness and the number of the capsules with abnormal wall thickness in the batches, calculating the proportion of the abnormal capsules in the total batch amount, evaluating the abnormal degree of the batches, and dividing early warning levels according to a set threshold value. And the response capability to production abnormity is improved.

The invention discloses a kidney lesion image analysis method and system based on a KTFE-YOLO model. The method comprises the steps of obtaining training data and processing the training data; carrying out lesion labeling on a kidney image and storing the lesion labeling as a YOLO format label; a KTFE-YOLO model is constructed: on the basis of a YOLOv11n-seg architecture, C3k2 of a sixth layer and a eighth layer in backbone is replaced by a fuzzy edge enhancement module, and C2PSA of a tenth layer is replaced by a multi-scale convolution fusion attention mechanism module; constructing a bounding box loss function to regress bounding box parameters; based on a bounding box loss function, performing parameter optimization on the KTFE-YOLO model by using the training data; and carrying out classification and segmentation on the kidney images by using the optimized KTFE-YOLO model. According to the method, the segmentation performance of fuzzy boundaries, tumors with complex shape features and small-size targets is remarkably improved while the real-time detection speed is kept.

The invention provides an aluminum veneer special-shaped cutting track planning method and system based on visual localization, and relates to the technical field of visual localization, and the method comprises the steps: employing an anisotropic guide filter to carry out the preprocessing of an aluminum veneer image, and maintaining the edge features through multi-scale filtering and adaptive weight fusion; carrying out edge enhancement treatment by combining material characteristics; extracting contour features by using a non-uniform rational B-spline contour descriptor, wherein the density of control points and the local curvature are in an exponential relationship; and finally, a cutting track is generated based on the feature processing network, and accurate cutting control over the special-shaped contour of the aluminum veneer is achieved.

The invention discloses an intelligent chest CT image processing system which comprises an image acquisition module, a multi-scale filtering processing module, a frequency domain edge enhancement module, an edge fusion correction module, an enhancement loss design module, a chest CT image segmentation model establishment module and a chest CT image processing module. The invention belongs to the field of image processing, and particularly relates to an intelligent chest CT (computed tomography) image processing system, which is characterized in that differentiated processing is performed in different frequency spectrum sections according to noise characteristics, key points are guided to be reserved, and blood vessel walking is enhanced; through edge fusion correction, adaptive reinforcement of small-range nodules and large-range lesions is considered; the branch A is used for refining the nodule edge and artifact weak contrast, and the branch B is used for strengthening the lung lobe macroscopic consistency; the decoding layer inhibits a learned region, focuses a blood vessel and a lung parenchyma junction through interface focusing; the loss is introduced into boundary smoothing, the overall segmentation contour is optimized, and false positive burr-shaped noise is reduced; and the chest CT image processing accuracy is improved.

The invention discloses an SAR ship instance segmentation method for self-adaptive representation alignment, and belongs to the field of SAR image instance segmentation. According to the implementation method, the problems of semantic-structural feature mismatch, global-local feature extraction mismatch and cross-dataset scale generalization mismatch are solved through modular design by constructing an adaptive representation alignment network. The method comprises an edge-guided boundary optimization module, a context sensing module and a depth adaptive feature pyramid module. The boundary optimization module introduces edge enhancement information in the feature extraction process to improve the target boundary positioning precision; the context sensing module fuses a multi-path global attention mechanism in a deep feature stage, and the semantic discrimination capability is enhanced; the depth adaptive feature pyramid module adaptively selects the optimal feature fusion depth by constructing a multi-depth fusion path in combination with scale statistical information, and improves the cross-dataset robustness. According to the method, the segmentation precision of the SAR ship instance can be remarkably improved.

The invention relates to a multi-modal image matching method and system based on learning features and epipolar geometric constraints. The method comprises the following steps: carrying out edge enhancement processing on an input image through wavelet transform; extracting a multi-scale dense feature map based on the transformed convolutional neural network, and generating a feature descriptor with rotation and scale invariance in combination with principal direction normalization; adopting an FLANN algorithm and dynamic distance constraint to realize preliminary feature matching; and introducing a basic matrix construction and epipolar geometric consistency verification mechanism, and eliminating mismatching point pairs in combination with an RANSAC affine constraint model. According to the method, image enhancement, deep learning and geometric verification strategies are fused, the problems of radiation nonlinearity and geometric distortion caused by imaging mechanism differences among multi-modal images are effectively solved, the matching precision and robustness are improved, and the method is suitable for remote sensing application scenes such as optical-SAR registration, multi-source image fusion and earth surface change detection.

The invention discloses a feature reconstruction-based tiny target detection method and device, a server and a medium, and belongs to the technical field of target detection. Comprising the following steps: inputting a primary feature obtained by convolution of an image into a two-layer DRSS-Unit module for processing to obtain a shallow edge enhancement feature; inputting the shallow-layer edge enhancement features into an MGSP-Unit module for processing to obtain middle-layer semantic transition features; the middle-layer semantic transition features are input into an MGSP-Unit module to be processed, and deep-layer sparse semantic features are obtained; the deep sparse semantic features are input into an SDF-AIFI module to be processed, and frequency domain global enhancement features are obtained; inputting the frequency domain global enhancement feature, the shallow edge enhancement feature and the middle semantic transition feature into a CCFM module for feature reconstruction to obtain a reconstructed fusion feature; and inputting the reconstructed fusion feature into a decoder to obtain a tiny target detection result. By constructing a cascade feature reconstruction and enhancement module, multi-level features are gradually reconstructed and enhanced, and detection of a tiny target in a complex scene is realized.