25 results about "Binary segmentation" patented technology

This application provides a method and apparatus for three-dimensional surface reconstruction of medical images based on diffusion least squares, belonging to the field of medical image processing. The method includes: edge extraction based on a binary segmentation mask of a three-dimensional image to determine multiple edge voxels of the three-dimensional image; based on the edge voxels, selecting multiple sampling voxels from the voxels of the three-dimensional image; performing least squares fitting on multiple local coordinate data of the multiple sampling voxels to determine multiple fitting parameter values ​​and calculating the corresponding residual values; if the residual value is greater than the fitting error threshold, determining the global coordinate data of the edge voxels through data mapping based on the multiple fitting parameter values; and obtaining a three-dimensional surface mesh through Poisson surface reconstruction and mesh iterative deformation based on the multiple global coordinate data.

The application discloses a method, system, device and medium for recognizing point fluorescence signals in cell nuclei. The method comprises inputting a pre-processed tissue slice scanning image into a cell nucleus segmentation network model to obtain a binary segmentation mask image; determining a target cell nucleus mask based on the binary segmentation mask image; multiplying a point fluorescence signal channel with the target cell nucleus mask to obtain a multiplication feature result image; inputting the multiplication feature result image into a point fluorescence signal recognition model to obtain a signal point heat map and a preliminary segmentation mask; fusing the signal point heat map, the preliminary segmentation mask and the point fluorescence signal channel to obtain a fusion feature image; inputting the fusion feature image into a signal point instance segmentation model to obtain a plurality of single signal point instances; and determining a target ACD score of the tissue slice scanning image based on the plurality of single signal point instances. The application can improve the accuracy, stability and efficiency of recognizing point fluorescence signals in cell nuclei.

A method for online temperature measurement target localization of furnaces and kilns based on dual-light fusion includes the following steps: a. Simultaneously acquiring visible light image sequences and thermal imaging image sequences of the surface of a rotating furnace and kiln through a dual-light data acquisition module; b. Processing the visible light image sequences and thermal imaging image sequences to generate a dense non-rigid deformation field, and applying the non-rigid deformation field to obtain a pair of geometrically corrected and spatially precisely aligned corrected image pairs; c. Fusion processing of the corrected image pairs, real-time sensing and adaptation to the interference level of different modal signals by the on-site environment, dynamically adjusting the fusion strategy, and generating a binary segmentation mask identifying thermal anomaly regions on the furnace and kiln surface; d. Using a kinematically aware spatiotemporal graph Kalman network (K-STGKN), performing state tracking and future state prediction on the thermal anomaly targets identified in the segmentation mask. This invention achieves high-precision, highly robust, and predictive localization and analysis of thermal anomalies on the surface of rotating industrial furnaces and kilns.

This application relates to a virtual binocular speckle stereo matching method based on local gray-level plane binary segmentation. It pertains to the fields of computer vision, 3D measurement, and stereo vision, and includes the following steps: S1: image loading and parameter settings; S2: local gray-level plane binary segmentation; S3: disparity calculation and sub-pixel optimization based on Hamming distance; S4: disparity map post-processing; sub-pixel interpolation is performed based on the matching cost curve to obtain disparity values ​​with sub-pixel accuracy, resulting in an initial sub-pixel disparity map; S5: depth map calculation and effective value filtering: the initial sub-pixel disparity map is filtered, consistency checked, and outlier removed to obtain an optimized dense disparity map; finally, combined with the calibration parameters of the virtual binocular system, the dense disparity map is converted into a depth map, and the result is visualized. This method has the advantages of high robustness, high accuracy and efficiency, and strong versatility.

The application provides a method for distinguishing the color of sub-group tobacco leaves, applied to the technical field of computer image processing, and comprises the following steps: obtaining a color classification sample image set of sub-group tobacco leaves, performing binary segmentation on the color classification sample of the tobacco leaves, and obtaining a binary image tobacco leaf area; performing coordinate return on the binary image tobacco leaf area, and generating a color image tobacco leaf area; extracting a pixel point Lab value set of the color image tobacco leaf area, combining a pure color pixel point Lab value, and calculating a pixel point contrast set; performing interval division on the pixel point contrast set, and generating a contrast interval division result; traversing the division result, and calculating an interval pixel proportion set; according to the interval pixel proportion set, drawing a proportion threshold point line graph, and then performing color distinction on a to-be-tested tobacco leaf image according to a voting mechanism. The method solves the technical problems that there is no standardized method for distinguishing the color of sub-group tobacco leaves in the prior art, manual classification has low efficiency and low classification accuracy, and is highly subjective.

The present application relates to a kind of high-precision crack segmentation method based on double-flow vision basic model cooperation, belong to computer vision and deep learning technical field, it aims at solving the problem of poor generalization of existing crack segmentation model under complex background, high-resolution under the problem of missing detection of fine crack, large model fine-tuning cost is high.The infrastructure surface image collected by the present application is preprocessed into 1024×1024 pixel standard input, and the features are extracted using a double-flow encoder with parallel SAM3 and DINOv3 double branches;Freeze the training parameters of double backbone, embed the lightweight adapter to complete efficient parameter fine-tuning;Extract double-branch multi-scale features, complete multi-scale splicing and dimensionality reduction fusion after channel alignment;Output the pixel-level crack binary segmentation result through the lightweight decoder based on depth separable convolution.The present application considers segmentation accuracy, cross-domain generalization ability and training deployment cost, and meets the engineering landing needs of infrastructure structure health monitoring.

This application discloses a binocular pupil recognition method based on high-brightness reflection point anchoring and contour matching, belonging to the field of automatic pupil recognition technology. It solves the problem of low pupil recognition accuracy in the prior art under complex conditions such as eyelash occlusion, corner shadows, and partial eyelid occlusion. The method includes: first, locating high-brightness reflection points in the left and right images of the binocular system and using them as stable spatial anchor points; then, determining an adaptive segmentation threshold in the monocular view; dynamically delineating the region of interest (ROI) centered on the centroid coordinates of the high-brightness reflection points and performing binary segmentation; identifying a preliminary set of contours and filtering them based on multiple features such as contour size, area, aspect ratio, and relative positional relationship with the anchor points to obtain a reliable set of candidate contours; finally, matching the candidate contours in the left and right views and selecting the optimal pupil contour pair, thereby achieving accurate and robust pupil localization in the binocular view.

The present application relates to the technical field of image data processing, and discloses a corn kernel and embryo volume high-throughput visual analysis method and system based on deep learning, which can obtain at least one folder according to a file input path set by a user, and each folder includes a gray scale sequence obtained by computer tomography (CT) slicing of corn kernel groups in a horizontal direction. Based on a parallel pipeline processing mode, a set batch size and a deep learning segmentation model, the gray scale sequence in the folder is preprocessed and image segmented to obtain kernel binary segmentation mask sequences and embryo binary segmentation mask sequences, and then the kernel quantity and the volume related data of the kernel and the embryo are determined, and the related images and data are visually output. The present application can complete the processing of the high-throughput data of the CT slice sequence of the corn kernel, and the nondestructive measurement and visual display of the volume related data of the kernel and the embryo, and effectively improve the measurement efficiency.

A macro-scale outcrop rock fracture network structure quantitative characterization method and system, the characterization method comprising: arranging a physical scale at the rock slope outcrop to be measured, and shooting a rock slope outcrop image; inputting the image into a pre-trained semantic segmentation model, segmenting the fracture pixel region in the image, and outputting a binary segmentation mask image; taking the physical scale as a reference, calibrating the spatial scale of the rock slope outcrop image to be measured transmitted to the data processing terminal, and calculating the spatial resolution of the image; using the binary segmentation mask image and the spatial resolution to respectively extract statistical parameters of each fracture, extract fractal parameters of the fracture network, and extract topological parameters of the fracture network; integrating the extracted statistical parameters, fractal parameters and topological parameters, introducing a SHAP attribution model, and forming a digital file of the rock slope outcrop fracture network structure, and quantifying the importance ranking and correlation of each parameter in the fracture network formation process.

This invention discloses a method and system for sample augmentation and lightweight deep learning model construction in power distribution line defect identification. The method includes: performing adaptive contrast stretching preprocessing on the input image and stitching normalized coordinate grids to enhance input information; constructing a lightweight detection model with two branches, one based on segmentation and the other on row classification, where the two branches share a partial feature extraction network and are fused through a feature interaction module to simultaneously output pixel-level segmentation results and row classification prediction results; employing a feature-response distillation method based on channel attention to transfer knowledge to a more streamlined student model; using a binary segmentation map to filter instance embedded features and performing mean-shift clustering to obtain independent instance segmentation results, which are then fused with row classification predictions for verification and supplementation; finally, deploying the compressed model to an embedded device to achieve real-time defect identification and localization. This invention effectively improves the detection accuracy and efficiency of the model in embedded environments.

This invention provides a method and device for tower displacement detection based on laser point cloud tower top feature extraction. The method includes: performing binary segmentation on the acquired tower laser point cloud to achieve precise separation of the tower from the transmission line; extracting point cloud data from the tower top region from the segmented tower point cloud and removing interference from overhead ground wire point cloud to obtain a pure rigid structure point cloud at the tower top; extracting feature anchor points from the pure rigid structure point cloud at the tower top, and combining this with the symmetry characteristics of the tower structure to obtain stable feature anchor points that are spatially unique, distributed on the same horizontal reference plane, and matched with the symmetrical structure of the tower; processing the tower top feature anchor points from different measurements to calculate the spatial displacement of the tower top and the horizontal torsional deformation angle of the tower; classifying the risk level of the tower structure and generating tower displacement detection results. This invention integrates the use of tower top symmetry and other features for feature anchor point refinement, achieving centimeter-level multi-dimensional monitoring, and is particularly suitable for ultra-high voltage transmission lines.

This invention relates to a deep learning-based method for detecting steel structure corrosion, belonging to the field of computer vision technology. It includes the following steps: acquiring steel structure corrosion images from multiple scenes to construct a detection dataset; performing adaptive contrast enhancement and standardization on the images; obtaining an image patch set through multi-scale non-uniform region of interest sampling; constructing a detection model including an adaptive neighborhood feature aggregation encoder and a detail-semantic dual-branch decoding architecture; after training and optimization using a multi-task joint loss function, the input image to be detected can output a binary segmentation mask of the corrosion region and a corrosion level label. This invention offers high detection accuracy and strong adaptability to on-site environments, enabling efficient and non-destructive detection of steel structure corrosion, and meeting the maintenance and inspection needs of steel structure engineering.

The application discloses a palm outline detection method and device, a storage medium and equipment, and belongs to the palm vein recognition field.The method comprises the following steps: adaptively setting a segmentation threshold according to an input image; performing binary segmentation on the input image by using the adaptively set segmentation threshold to obtain a foreground region and a background region; and finding a palm boundary based on the palm region to obtain a palm outline.In the binary segmentation process, the segmentation threshold is adaptively set according to the input image, the result after the input image is segmented by using the adaptively set segmentation threshold is more accurate, and the accuracy of palm outline detection is improved.

The application discloses a flood influence range extraction method and system based on convolution and state space model, and belongs to the technical field of remote sensing image processing and disaster monitoring; the method comprises the following steps: acquiring remote sensing image data of a region to be detected, and performing pretreatment; inputting the pretreated remote sensing image into a multi-stage and multi-branch hierarchical feature extraction encoder; utilizing channel attention and spatial attention mechanisms of the hierarchical feature extraction encoder to dynamically weight and fuse local and global features of different scales; inputting the fused multi-scale feature maps into a decoder for upsampling, and finally outputting a pixel-level binary segmentation mask of the flood influence range. The application has the advantages of high calculation efficiency, accurate edge positioning and strong anti-interference capability, and is suitable for actual businesses such as emergency monitoring, disaster assessment and post-disaster reconstruction.

This invention discloses a surgical instrument segmentation method combining YOLOv8 and SAM2, belonging to the field of computer-aided surgical perception and medical image processing technology. This invention introduces a YOLOv8 detection and geometric state maintenance mechanism to ensure stable and accurate spatial observation information during temporal segmentation. First, a surgical video instrument segmentation dataset is constructed to form a binary segmentation task. Then, a joint inference framework consisting of a YOLOv8 detection module and a SAM2 video segmentation module is constructed, using frame-by-frame detection results to provide stable spatial cues and observation information for SAM2. Furthermore, through lightweight geometric state maintenance, geometric consistency matching, smooth state updates, and missing state management mechanisms, the state of the target in consecutive frames is constrained. Finally, the optimized instrument segmentation result is output. This invention can effectively improve the instrument segmentation accuracy in complex surgical video scenarios and has good clinical application prospects.

The present application relates to the technical field of medical image processing, and discloses a kind of enhanced SAM's child dermatoscope image segmentation method, comprising: collecting child dermatoscope image and repairing circular reflection area, through adaptive light correction and skin color standardization processing to enhance lesion visibility and eliminate individual color difference;Image is divided into non-overlapping block sequence, projection and fusion learnable position coding to obtain feature sequence;Through double-branch parallel extraction global semantics and local boundary texture features, using spatial perception gate mechanism to adaptively fuse double-branch features;Double-layer residual connection is carried out to the fusion features and stacked multiple layers to form deep encoder, obtain multi-scale feature pyramid;Layer by layer up-sampling and fusing each layer feature, through multi-scale boundary refinement network to deal with fuzzy boundary to obtain binary segmentation mask.The present application can improve the segmentation accuracy and computational efficiency of fuzzy boundary lesions in child dermatoscope image.

An image decoding method is described for: determining if the height and / or width of a current encoding unit is greater than a predetermined size; acquiring, from a bitstream, a second encoded block flag indicating whether a luminance component block included in at least one transformation unit includes at least one transformation coefficient in the bitstream, based on whether to segment the current encoding unit into transformation units; acquiring a residual signal from the luminance component block included in the at least one transformation unit, based on the second encoded block flag; reconstructing the current encoding unit based on the residual signal; and reconstructing a current image that includes the current encoding unit, based on the reconstructed current encoding unit.A segmentation form mode indicates at least one of segmentation, segmentation direction, and segmentation type, and the segmentation type can indicate one of binary segmentation, triple segmentation, and quad segmentation.

This invention discloses a portable intelligent wound measurement method and system based on the fusion of Hough circle transform and deep network, belonging to the field of artificial intelligence medical image processing. The method includes the following steps: acquiring wound images containing circular calibration objects and ensuring standardized image capture through a pose awareness mechanism; performing feature enhancement and preprocessing on the images; anchoring the calibration objects using an adaptive parameter Hough circle transform algorithm and obtaining pixel radii to establish a spatial scale; inputting the images into a pre-trained U-Net++ deep semantic segmentation network, extracting features using dense skip connections, and outputting a binary segmentation mask of the wound surface; finally, retrieving the true area and size based on the mask features and the scale. This invention uses a deep network to extract the wound surface and combines it with Hough circle transformation for scale fixation and pixel calculation, effectively improving the segmentation accuracy and measurement accuracy of complex wounds, achieving automated quantification of wound size and area, reducing manual measurement errors, and improving the efficiency of wound assessment and management.

This invention provides a parameter-freezing fuzzy attention method for 3D abdominal organ segmentation, relating to the field of intelligent medical image processing technology. The technical solution includes the following steps: S1: Obtaining 3D abdominal organ medical image data and preprocessing the data; S2: Constructing an architecture-based image encoder; S3: Inputting the preprocessed 3D abdominal organ image data into the image encoder and executing a parameter-freezing strategy; S4: Feeding the obtained deep semantic features into a fuzzy spatial attention module; S5: Obtaining geometric cue information for the target region; S6: Stacking and reconstructing consecutively generated two-dimensional binary segmentation masks according to the spatial index order of the original medical image, outputting the final 3D abdominal organ segmentation volume data. This invention reduces training memory usage through parameter freezing.

The application discloses an air traffic control primary radar target detection method based on end-to-end, relates to the field of air traffic control primary radar, and solves the problem that the prior art relies on artificial experience and an existing deep learning model is difficult to effectively utilize radar space-time information. The method comprises the following steps: acquiring and pre-processing real-time radar data to obtain a real-time matrix, combining the real-time matrix with historical matrices of two continuous time points into a space-time three-dimensional data block; inputting the data block into a trained target segmentation model to obtain a binary segmentation graph; extracting a multi-dimensional feature vector of each connected region through connected domain analysis; inputting the multi-dimensional feature vector into a trained target classification model to obtain a probability value of each set belonging to a real target; and finally outputting a target track list through threshold screening, non-maximum suppression and coordinate mapping. The application realizes end-to-end automation of the detection process, can adapt to complex environments, significantly improves detection performance and reduces dependence on artificial experience.

This invention discloses a medical image segmentation method, apparatus, storage medium, and device, belonging to the field of medical image segmentation technology. The method includes acquiring a liver ultrasound image and inputting it into a trained backbone network to generate initial feature maps at several levels; aggregating the initial feature maps from levels 2 to 1 into a level 1 segmentation map; restoring the resolution of the initial feature maps from levels 1 to 1 to the resolution of the liver ultrasound image to obtain corresponding optimized feature maps; inputting the level 1 segmentation map and optimized feature maps into a trained segmentation model, fusing the multi-scale information of each optimized feature map into the level 1 segmentation map in a reverse additive attention module to obtain a fusion result, and then converting it into a binary segmentation mask with the same size as the liver ultrasound image. This invention improves the clarity and accuracy of boundary segmentation through a reverse additive attention module and enhances the accuracy of multi-scale tumor target detection by combining feature aggregation and resolution restoration.

Techniques are disclosed herein for implementing a novel, low latency, guidance map-free video matting system, e.g., for use in extended reality (XR) platforms. The techniques may be designed to work with low resolution auxiliary inputs (e.g., binary segmentation masks) and to generate alpha mattes (e.g., alpha mattes configured to segment out any object(s) of interest, such as human hands, from a captured image) in near real-time and in a computationally efficient manner. Further, in a domain-specific setting, the system can function on a captured image stream alone, i.e., it would not require any auxiliary inputs, thereby reducing computational costs—without compromising on visual quality and user comfort. Once an alpha matte has been generated, various alpha-aware graphical processing operations may be performed on the captured images according to the generated alpha mattes (e.g., background replacement operations, synthetic shallow depth of field (SDOF) rendering operations, and / or various XR environment rendering operations).

This invention relates to a vision-based intelligent assessment method for recyclable waste pollution levels, comprising the following steps: S1: acquiring an original video frame dataset; S2: preprocessing the acquired original video frame dataset; S3: performing recyclable waste target detection and instance segmentation based on the preprocessed image frame dataset to obtain an instance set; S4: cropping the ROI from the entire image for each instance using bounding box parameters, and using a binary segmentation mask for background suppression to obtain a standard input set for each object; S5: using hierarchical recognition based on the standard input set for each object to obtain the material and fine-grained category of each object; S6: performing pollution segmentation under object mask constraints, outputting a mask categorized by pollution type and calculating the area proportion of each pollution type; S7: calculating the comprehensive pollution level based on the material and fine-grained category of each object and the area proportion of each pollution type to obtain a pollution score. This invention significantly improves the efficiency and reliability of recyclable waste pollution level assessment.

The application provides a kind of based on image recognition's degree of needle erosion evaluation method and device, it is applied to image recognition technical field, above-mentioned method includes: obtaining the needle picture to be evaluated;The needle picture is input to the trained erosion area identification model, and the classification probability graph output by the erosion area identification model is obtained;The classification probability graph is carried out binary segmentation processing, and the erosion area is obtained;Determine the pixel point number of the needle cone area corresponding to the needle picture in the erosion area;The pixel point number is compared with the preset erosion pixel point number threshold, and the degree of needle erosion of the needle picture is determined. Through the application, the detection efficiency of needle erosion can be improved, and false judgment and missed judgment caused by human subjective factors can be avoided.

This invention relates to a deep learning-based system and method for detecting landing points on flat ground for unmanned aerial vehicles (UAVs). The system includes: a ground segmentation module for acquiring airborne images captured by the UAV, performing inference on the images, and outputting a binary segmentation mask representing flat, landable, and non-landable areas; a pixel radius calculation module for calculating the target pixel radius corresponding to the actual landing radius in the current airborne image; a landing point detection module for filtering one or more candidate landing points that meet the safety radius constraint based on the binary segmentation mask and the target pixel radius; and an output module for outputting the detection results containing the candidate landing points. This system solves the problems of scene adaptation, safety radius consistency, and decision reliability in UAV vertical-view scenarios, achieving real-time detection of reliable landing points that meet safety radius requirements.