32 results about "Texture model" patented technology

The invention discloses a metal product defect detection method and system based on image recognition. The method comprises the following steps: firstly, executing reflection disturbance digestion processing on a surface image of a to-be-detected metal product to obtain a reflection digestion image; obtaining surface reference texture features of the defect-free metal product, and generating a reference texture model on the basis of a texture distribution rule, a gray average value and texture continuity; performing defect texture gradient separation on the reflection resolution image based on a reference texture model, positioning an abnormal region, and performing segmentation to obtain a suspected defect texture region; performing boundary pixel reconstruction on the suspected defect texture region to obtain a defect texture reconstruction region; obtaining a target defect texture region through local variance enhancement and neighborhood correlation analysis; and finally, the overexposure area is positioned, gray inverse stretching processing is performed, abnormal pixel group feature clustering is performed on the preprocessed defect area, and then a surface defect detection result is output, so that the precision and accuracy of metal product surface defect detection are improved, and the false detection rate is reduced.

The invention discloses a Hololens-oriented single-view texture-free object pose estimation method, belongs to the technical field of augmented reality, and solves the problem that an existing Hololens pose estimation method cannot perform accurate pose estimation on a texture-free model through single-view data under the conditions of no marker and no pre-training. According to the method, multi-modal data acquisition, self-supervised segmentation, self-adaptive point cloud denoising and geometrically-driven pose optimization are integrated, and accurate pose estimation is carried out on a texture-free object only through single-view-angle data under the condition that no marker exists and pre-training is not carried out. Through multi-modal data and spatial transformation, a point cloud reconstructed by a Depth image is mapped to a visual angle of an RGB camera, so that the point cloud is conveniently filtered by using a target mask obtained on the RGB image. Meanwhile, rendering templates under multiple visual angles are generated by rendering the CAD model, obstacles caused by no texture of the target model are avoided, meanwhile, the CAD model does not depend on pre-training, and the zero sample learning ability is provided.

The application relates to the technical field of image processing, in particular to a thermal imaging visualization method and system for energy efficiency analysis of a roasting process. The method comprises the following steps: acquiring a thermal imaging image of a roasting kiln body surface, and calculating local complexity indexes and local anisotropy indexes of each pixel point in the thermal imaging image; extracting a multi-scale and multi-direction texture feature set of each pixel point in the thermal imaging image; calculating scale weights of each scale and direction weights of each direction, and performing weighted fusion on the texture feature set to obtain a fusion feature vector of each pixel point; establishing a normal texture model based on the fusion feature vector, and calculating an abnormal score image of the thermal imaging image to determine an abnormal heat loss area and perform visualization. Through the technical scheme, different scale texture changes can be effectively identified, and the detection accuracy of the abnormal heat loss area is improved.

This invention provides a weight training method and system for generating geological lithological texture models based on LoCon, belonging to the field of geological lithological texture generation technology. It employs an LDM model as the large model framework for weight training and texture generation. Convolutional layers are integrated into the convolutional weights of the residual blocks in the U-Net model within the LoCon model. The low-rank matrix of the LoCon model is added to the pre-trained weight matrix in the U-Net model to construct the generated geological lithological texture model. The model is trained using a training set. The trained weights are output, loaded into the large model framework for inference, and the training parameters are adjusted based on the inference results. Training is repeated until the preset inference result requirements are met. An API is provided to create an interface between the LDM large model framework and the trained weight matrix. This fills a gap in the geological industry's material library and lowers the technical threshold for geologists to participate in digitization.

The invention relates to the technical field of three-dimensional visualization, in particular to a three-dimensional visualization collaborative design method under a complex terrain, and solves the technical problem of poor three-dimensional modeling effect in the prior art. The method comprises the following steps: acquiring point cloud data of a target object scanned by a plurality of scanning sites, and acquiring a multi-view image of the target object acquired by image acquisition equipment; the multi-view image comprises a plurality of target images collected by the target object in different view directions; identifying an abnormal brightness area from the multi-view image, and repairing the abnormal brightness area to generate a repaired multi-view image; and performing three-dimensional reconstruction and image texture mapping based on the repaired multi-view image and the point cloud data to generate a three-dimensional scene texture model.

The invention relates to the technical field of image processing, in particular to a textile material surface defect detection system based on machine vision, the system comprises a processor and a memory, the processor executes a computer program of the memory to realize the following steps: obtaining a spectrogram of a textile material to be detected and the amplitude of each frequency component in the spectrogram, dividing the spectrogram into at least two initial sub-regions, acquiring a peak value prominence degree, an amplitude distribution obvious degree and a dominant frequency existence index of any initial sub-region, and according to the peak value prominence degree, the amplitude distribution obvious degree and the dominant frequency existence index of any initial sub-region, determining the dominant frequency existence index of any initial sub-region; adjusting any initial sub-region to obtain a target sub-region; and obtaining the target sub-region corresponding to each initial sub-region, obtaining the dominant frequency in each target sub-region, constructing the normal texture model, and carrying out surface defect detection on the to-be-woven material, so that the detection result of the surface defect of the textile material is improved.

The invention relates to the technical field of man-machine interaction, and discloses a multi-modal AI and emotional intention recognition three-dimensional content real-time generation method, which integrates a multi-modal AI collaborative generation module, a dynamic emotional intention recognition module, a three-dimensional content real-time generation module and a holographic software and hardware collaborative module, the method comprises the following steps: realizing an immersive holographic interaction and multi-modal AI collaborative generation module, constructing a unified semantic space by adopting CLIP + VATT, analyzing an instruction by adopting LLM to generate parameters, generating 2D content by adopting Diffusion, converting NeRF into a textured 3D model in real time, and realizing sketch / semantic driven parameterization generation by combining ControlNet and LoRA. According to the scheme of the invention, the breakthrough efficiency can be improved, and the AI driven automatic generation technology is realized; a traditional 3D content production process which needs to be completed in several days can be compressed to be completed in several minutes by means of collaborative optimization of models such as Point-E, NeRF and Diffusion and combining ControlNet structured control and LoRA low-rank fine tuning, and the modeling efficiency is integrally improved.

Embodiments of this application provide a method for generating a three-dimensional texture model of a city based on composite data, and a device. The method includes: obtaining satellite remote sensing data of a target region, where the satellite remote sensing data is used to determine a three-dimensional framework model and spatial coordinates of a to-be-measured object; obtaining video data within a preset range of the spatial coordinates of the to-be-measured object, where the video data is used to determine a surface texture of the to-be-measured object; and generating a three-dimensional texture model of the to-be-measured object based on the surface texture and the three-dimensional framework model of the to-be-measured object, where the three-dimensional texture model of the to-be-measured object is used to generate a three-dimensional texture model of the target region. The satellite remote sensing data and the video data within the preset range of the spatial coordinates of the to-be-measured object are obtained at low costs, so that costs and difficulty of constructing the three-dimensional texture model of the city are reduced, and costs and difficulty of updating the three-dimensional texture model of the city are reduced.

This application discloses a method and system for detecting defects in metal products based on image recognition. The method first performs reflection perturbation elimination processing on the surface image of the metal product to be detected to obtain a reflection elimination image; then, it acquires the surface reference texture features of the defect-free metal product, and generates a reference texture model based on texture distribution patterns, gray-level mean, and texture continuity; next, it performs defect texture gradient separation on the reflection elimination image based on the reference texture model, locates abnormal regions, and segments them to obtain suspected defect texture regions; then, it performs boundary pixel reconstruction on the suspected defect texture regions to obtain defect texture reconstruction regions; the target defect texture region is obtained through local variance enhancement and neighborhood correlation analysis; finally, it locates overexposed regions and performs gray-level inverse stretching processing, and outputs the surface defect detection results after clustering abnormal pixel group features of the preprocessed defect regions, thereby improving the accuracy and precision of surface defect detection of metal products and reducing the false detection rate.

The invention discloses a palm vein recognition method and system based on multispectral imaging and deep learning and a storage medium, and relates to the field of biological feature recognition, and the method comprises the steps: obtaining the multi-dimensional information of a palm through the synchronous collection of near-infrared, short-wave infrared and visible light wave band images, and then carrying out the adaptive weighted fusion and enhancement processing of a multispectral image, thereby achieving the recognition of the palm vein. In the core recognition step, a double-branch deep learning model is adopted, one branch extracts static structure features from the vein feature map, the other branch analyzes blood flow dynamic features from a time sequence image, and finally the static structure features and the blood flow dynamic features are fused for integrated living body detection and identity recognition. Through three-dimensional vein texture modeling and a dynamic living body detection anti-counterfeiting mechanism, faking attacks of photos, silica gel molds and the like are completely eradicated, and the problems that in a complex environment, performance of a traditional method is reduced, faking attacks are likely to happen, and speed and precision are difficult to consider at the same time are effectively solved.

This invention relates to a texture primitive extraction method for tire X-ray images, addressing the challenge of accurately obtaining pixel-by-pixel texture information at the individual cord level while suppressing background interference. It falls under the field of computer vision and image processing technology. The method combines frequency domain analysis to obtain texture direction and spacing, utilizes this information to construct a mesh mask, and further combines background point extraction with real background reconstruction to filter out specific background regions in the original image. Under directional constraints, the remaining texture mesh is continuously tracked to obtain pixel-by-pixel texture information, thus achieving texture primitive extraction. This provides a more reliable foundation for subsequent pathological detection, structural analysis, and cord-level texture modeling.

The application relates to a crowd-sourced unordered image auxiliary-based urban scene reconstruction method and device and a storage medium, wherein the method comprises the following steps: acquiring laser radar data to form basic point cloud data; acquiring crowd-sourced unordered images, estimating a depth map by using a depth estimation network, and obtaining auxiliary point cloud data; fusing the basic point cloud data and the auxiliary point cloud data to obtain geometric information; acquiring multispectral images and panchromatic images obtained by satellite remote sensing, determining ground object color information based on the multispectral images, and determining a ground surface texture model and a ground object model based on the panchromatic images; performing data preprocessing on the crowd-sourced unordered images to obtain a texture image, combining the ground object model and the ground object color information to obtain a ground object texture model; determining texture information based on the ground object texture model and the ground surface texture model; and realizing urban scene reconstruction by using an automatic reconstruction solution based on the geometric information and the texture information. Compared with the prior art, the application has the advantages of high restoration degree and high reconstruction precision.

Aspects of this technical solution can, according to a multitude of cameras oriented towards the surface of a three-dimensional (3D) model with a surface containing a two-dimensional (2D) texture model, receive input according to corresponding views from the multitude of cameras of the 2D texture model on the surface of the 3D model and, according to the input and according to a model configured to generate a two-dimensional (2D) image, generate an output containing a 2D texture for the 3D model, the output being generated in response to receiving a specification of the 3D model and the 2D texture.

This application discloses a method for rendering geological attribute meshes, including: obtaining geometric outer surface parameters; constructing a hollow model shell based on the geometric outer surface parameters; obtaining the coordinates of the hollow model shell; performing attribute mapping on the hollow model shell based on the coordinates of the hollow model shell to obtain a textured model; parsing the textured model to determine the geological body mesh on the textured model; merging the faces of geological body meshes with similar attributes on the textured model to obtain a low-precision model; and rendering the low-precision model to obtain a high-precision textured model. This method effectively reduces the number of geometric shapes such as triangles, simplifying the model, reducing the amount of rendering data, reducing the computational load, improving rendering efficiency, and resulting in better rendering effects. Because the attribute mapping uses high-precision attributes, it effectively ensures that the texture details and attribute information of the geological bodies are preserved.

The invention provides a lightweight image super-resolution method based on super-pixel guidance, and the method comprises the steps: carrying out the processing of a low-resolution image through SLIC, so as to obtain a super-pixel label graph; performing convolution mapping on the low-resolution image by using lightweight convolution to obtain a shallow feature map; taking the shallow feature map as an input feature map, and inputting the input feature map into a feature enhancement block to obtain an output feature map of a middle feature enhancement block; taking the output feature map of the middle feature enhancement block as the input feature map of the next feature enhancement block, and obtaining a deep feature map after stacking a predetermined number of feature enhancement blocks; obtaining a final feature map by using the deep-layer feature map and the shallow-layer feature map; obtaining a high-resolution image through the final feature map and the low-resolution image; according to the method, 'region-level context ', 'structure-sensitive sparse aggregation' and'fine-grained texture modeling 'are unified into the same lightweight module, so that both structure maintenance and detail recovery are considered at lower cost.

The invention relates to the technical field of artificial intelligence and computer vision, in particular to an OCR (Optical Character Recognition) method based on large model enhancement, which comprises the following steps of: performing multi-direction differential operation and smooth processing on an input image to generate a direction texture feature map; performing multi-scale feature extraction and adaptive weighted fusion to generate a multi-scale aggregation feature map; calculating energy statistics to generate a space gating weight, and performing weighted enhancement on the aggregated feature map to obtain a space enhanced feature map; performing feature transformation on the spatial enhancement feature map and the direction texture feature map, and fusing deep and shallow layer features to generate uniform feature representation; and reconstructing a character form in a visual branch based on the representation, performing semantic reasoning in a language branch, and fusing double-branch results to obtain a target recognition result. According to the invention, through direction texture modeling, adaptive multi-scale fusion and vision-language double-branch cooperation, the accuracy and robustness of OCR recognition in a complex scene are significantly improved.

The application discloses a palm vein recognition method and system based on multispectral imaging and deep learning and a storage medium, relates to the field of biometric recognition, and comprises the following steps: first, acquiring multi-dimensional information of a palm by synchronously collecting near-infrared, short-wave infrared and visible light band images; then, adaptively weighting and fusing the multispectral images and performing enhancement processing to generate a high-quality vein feature map; and finally, adopting a double-branch deep learning model for core recognition, one branch extracting static structural features from the vein feature map and the other branch analyzing blood flow dynamic features from time-series images, and finally fusing the two for integrated living body detection and identity recognition. The application adopts three-dimensional vein texture modeling and dynamic living body detection anti-forgery mechanism, completely eliminates photo, silicone mold and other forgery attacks, and effectively solves the problems of performance decline in a complex environment, vulnerability to forgery attacks and difficulty in balancing speed and accuracy of the traditional method.

The present application relates to the technical field of image recognition, and particularly relates to a risk detection method, equipment and medium for a mountainous power transmission line driven by unmanned aerial vehicle tilt photography data. The method comprises the following steps: acquiring image data collected by a tilt photography camera and generating a dense point cloud model; constructing a current state digital ground surface model and a real scene three-dimensional texture model based on the dense point cloud model, and extracting a tower foundation area by using a spatial semantic segmentation method combined with image recognition; performing change detection on the current digital ground surface model and a historical digital ground surface model, and performing auxiliary observation on a deformation area by combining the current and historical real scene three-dimensional texture models to determine a suspected landslide deformation area; and performing risk level evaluation on the suspected landslide deformation area in combination with the tower foundation area. The method starts from four aspects of data acquisition, three-dimensional modeling, time series analysis and intelligent recognition, overcomes the limitations of traditional methods in engineering applicability, processing precision and dynamic response, and provides support for power transmission line disaster monitoring.

Techniques are described with respect to a system, method, and computer program product for visualizing weather in a digital environment. An associated method includes; receiving a plurality of weather data; converting the plurality of weather data for a texture model associated with a plurality of virtual objects associated with the digital environment; and visualizing at least one derivative of the conversion in the digital environment based on the model.

A method is performed at a moveable scanner with one or more optical sensors. The method includes scanning, using the moveable scanner, an object having a surface. The scanning generates color data from a plurality of orientations of the moveable scanner with respect to the object. The method further includes generating, using at least the color data, a pixel map of the surface of the object, the pixel map including, for each respective pixel of a plurality of pixels: a color value of a corresponding point on the surface of the object; and a value for a non-color property of the corresponding point on the surface of the object.

This invention relates to the fields of artificial intelligence and computer vision, and particularly to an OCR recognition method based on large model enhancement. This method generates a directional texture feature map by performing multi-directional difference operations and smoothing on the input image; extracts multi-scale features from the map and adaptively weights and fuses them to generate a multi-scale aggregated feature map; calculates energy statistics to generate spatial gating weights, and weights and enhances the aggregated feature map to obtain a spatially enhanced feature map; performs feature transformation on the spatially enhanced feature map and the directional texture feature map, and fuses deep and shallow features to generate a unified feature representation; based on this representation, reconstructs character morphology in the visual branch and performs semantic reasoning in the language branch, fusing the results of the two branches to obtain the target recognition result. This invention significantly improves the accuracy and robustness of OCR recognition in complex scenes through directional texture modeling, adaptive multi-scale fusion, and visual-language dual-branch collaboration.

The invention relates to the technical field of image recognition, in particular to a mountain area power transmission line risk detection method driven by unmanned aerial vehicle oblique photography data, equipment and a medium. The method comprises the following steps: acquiring image data acquired by an oblique photogrammetry camera, and generating a dense point cloud model; constructing a digital earth surface model and a live-action three-dimensional texture model of a current state based on a dense point cloud model, and extracting a tower basic region by combining a spatial semantic segmentation method with image recognition; performing change detection on the current digital earth surface model and the historical digital earth surface model, performing auxiliary observation on the deformation area in combination with the current and historical live-action three-dimensional texture models, and determining a suspected landslide deformation area; and carrying out risk grade evaluation on the suspected landslide deformation area in combination with the tower foundation area. Starting from four aspects of data acquisition, three-dimensional modeling, time sequence analysis and intelligent identification, the limitation of a traditional method in engineering applicability, processing precision and dynamic response is overcome, and support is provided for power transmission line disaster monitoring.

The application relates to the technical field of three-dimensional modeling, and discloses a three-dimensional scene establishment method and system based on virtual and real scripts. The method comprises the following steps: performing deep semantic analysis according to a semantic description text, extracting emotional atmosphere and narrative logic features, obtaining a semantic feature vector, and matching and screening visual elements in a pre-stored visual element resource library to obtain a visual element list; generating spatial position and layout relationship of the visual elements according to the visual element list and a user interaction log, obtaining a layout parameter set, establishing a topological connection relationship, and generating a scene skeleton structure; performing mapping and synthesis of surface textures according to the scene skeleton structure to obtain a texture model; performing iterative adjustment of light and shadow parameters according to the texture model to obtain rendering configuration parameters; and performing three-dimensional scene drawing and pixel correction according to the rendering configuration parameters to obtain three-dimensional scene data. The method can generate a three-dimensional scene with high consistency of emotion and narrative logic according to semantic description.

The invention relates to the technical field of three-dimensional modeling, and discloses a three-dimensional scene establishment method and system based on virtual and real scripts. The method comprises the following steps: performing deep semantic analysis according to a semantic description text, extracting emotion atmosphere and narrative logic features to obtain semantic feature vectors, and matching and screening the semantic feature vectors with visual elements in a pre-stored visual element resource library to obtain a visual element list; according to the visual element list and the user interaction log, generating a spatial position and layout relationship of the visual elements, obtaining a layout parameter set, establishing a topological connection relationship, and generating a scene skeleton structure; mapping and synthesizing surface textures according to the scene skeleton structure to obtain a texture model; performing light and shadow parameter iterative adjustment according to the texture model to obtain rendering configuration parameters; and performing three-dimensional scene drawing and pixel correction according to the rendering configuration parameters to obtain three-dimensional scene data. According to the method, a three-dimensional scene with emotion highly consistent with narrative logic is generated according to semantic description.

The invention discloses an ancient tomb archaeological site modeling method, and relates to the technical field of image analysis, and the method comprises the steps: planning a three-dimensional laser scanning point and a scanning path for an ancient tomb site, and setting a target; performing three-dimensional laser scanning according to the scanning path; carrying out data preprocessing on the collected point cloud data, and splicing to obtain an initial point cloud model of the coffin chamber; planning an image acquisition range and a shooting angle for a key object in the coffin chamber, and performing multi-view image acquisition on the key object; carrying out image preprocessing on the key feature image, and importing the key feature image into three-dimensional reconstruction software to generate a dense point cloud and texture model; and fusing the initial point cloud model of the tomb chamber with the dense point cloud and texture model to obtain an ancient tomb archaeological site three-dimensional model. According to the technical scheme, the whole process from data collection to model generation is optimized, efficiency can be guaranteed, details can be highlighted, the precision and integrity of the model are comprehensively improved, and powerful support is provided for archaeological work.

The invention discloses a tight sandstone gas reservoir dynamic reserve calculation method and device, and the method comprises the steps: carrying out the initialization of a gas reservoir numerical simulation model based on a three-dimensional texture model and gas reservoir data, and completing the building of the gas reservoir numerical simulation model; carrying out production history fitting on the gas reservoir numerical simulation model to realize verification of the gas reservoir numerical simulation model; performing production dynamic prediction and single-well dynamic reserve time-sharing calculation by using the gas reservoir numerical simulation model to obtain single-well dynamic reserves at a certain moment; and performing data processing and final value determination on the single-well dynamic reserves at different moments to obtain the single-well dynamic reserves at any moment. The method can solve the problems that the calculation of the tight sandstone gas reservoir single well dynamic reserves is greatly influenced by production time and is greatly influenced by the flow state and the like.

The invention relates to the field of data processing, and discloses a three-dimensional reconstruction method and system in a night low-illumination scene, a terminal and a storage medium, and the method comprises the steps: constructing a three-dimensional geometric texture model, obtaining a night low-illumination simulation scene based on the texture model, simulating a low-altitude collection track of an unmanned aerial vehicle in the simulation scene, and obtaining a low-altitude collection track of the unmanned aerial vehicle; acquiring night real data and day real data in a real city real scene; constructing a target reconstruction data set based on the texture model, the simulation scene, the unmanned aerial vehicle low-altitude acquisition trajectory, the night real data and the daytime real data; constructing an enhanced network model, and training the enhanced network model by using the target reconstruction data set and the low-illumination image data set to obtain a target enhanced network model; and obtaining a night low-illumination image to be processed, inputting the night low-illumination image into the target enhancement network model to obtain an enhanced image, and reconstructing a 3D point cloud based on the enhanced image. According to the invention, three-dimensional reconstruction application in a city night scene can be efficiently realized.

This invention proposes a lightweight image super-resolution method based on superpixel guidance. The method includes: processing a low-resolution image using SLIC to obtain a superpixel label map; performing a convolutional mapping on the low-resolution image using lightweight convolution to obtain a shallow feature map; using the shallow feature map as input to a feature enhancement block to obtain the output feature map of an intermediate feature enhancement block; using the output feature map of the intermediate feature enhancement block as input to the next feature enhancement block, and stacking a predetermined number of feature enhancement blocks to obtain a deep feature map; using the deep feature map and the shallow feature map to obtain a final feature map; and obtaining a high-resolution image using the final feature map and the low-resolution image. This invention unifies "region-level context," "structure-sensitive sparse aggregation," and "fine-grained texture modeling" into a single lightweight module, achieving both structure preservation and detail recovery at a lower cost.

The application discloses a medical image segmentation method based on boundary constraint, S1. Obtain medical image amplitude spectrum and medical image phase spectrum; S2. Form structure perception frequency domain disturbance result; S3. Obtain frequency domain enhanced medical image; S4. Calculate structure perception medical image boundary enhancement attention output feature; S5. Global relationship modeling and local texture modeling are carried out on the frequency domain enhanced medical image sequence; S6. Generate edge heat map; S7. Bidirectional gate interaction is carried out on the edge heat map and linear self-attention encoder output by adopting a dual-domain interactive fusion strategy, a dual-domain interactive fusion feature map is generated, and the dual-domain interactive fusion feature map is up-sampled and convolution-processed in the decoder, and a boundary enhancement segmentation feature map is output. The application effectively improves the structure perception capability of the model in the medical image scene with fuzzy anatomical structure boundary and low texture contrast.

This invention provides a method and apparatus for editing digital humans based on neural texture and 3D Gaussian models. The method includes: modeling the expression, posture, and appearance of the digital human using a 3D Gaussian splash model based on video data of a target object's motion sequence, obtaining a Gaussian model of the digital human; modeling the expression, posture, and appearance of the digital human using a neural texture model, obtaining a neural texture model of the digital human; pruning the Gaussian model, and performing a blended rendering based on the rendered image output by the pruned Gaussian model and the rendered image output by the neural texture model of the digital human, obtaining a blended rendered image; and optimizing the Gaussian model of the digital human based on the blended rendered image and the motion sequence video data, obtaining an editable digital human model. The method of this invention reduces the storage overhead and structural complexity of the digital human model, and improves the high-frequency details and editing efficiency of the digital human model's skin surface.