57 results about "Texture representation" patented technology

In one implementation, we propose to provide signaling to inform applications about a representation of a texture as a linear combination of textures. The representation of the proposed format is reconstructed given at least one texture target with an associated target weight. This representation facilitates the transfer of parametric meshes with parametric textures.

The invention relates to a light-weight low-light image enhancement method based on illumination iterative adjustment, and the method comprises the steps: carrying out the HVI color space transformation of a to-be-processed image, obtaining an HVI image, separating an illumination intensity component, obtaining an illumination intensity feature map, and carrying out the convolution processing of the HVI image, and obtaining a shallow feature map; feature fusion and brightness adjustment are carried out on the illumination intensity feature map and the shallow layer feature map to obtain a depth feature map, up-sampling is carried out step by step through two layers of decoders to recover the size, channel adjustment is carried out on each layer of decoder by adopting convolution of a specified size, and up-sampling is realized through bilinear interpolation; a Mama-based structure refining module is introduced into the tail end of each layer of decoder to improve the structure restoration capability, the number of channels of HVI features output by the last layer of decoder is adjusted to a specified value through a convolution of a specified size, residual connection is carried out on the HVI features and an initial HVI image, an enhanced HVI image is obtained, and a restored image is obtained through inverse HVI transformation. Adaptive enhancement is carried out under various low light conditions, multi-directional spatial context aggregation is realized, and texture representation is improved.

The invention discloses a stylized image guided three-dimensional model texture generation method, computer equipment, a storage medium and a program product. The stylized image guided three-dimensional model texture generation method is a pipeline based on a diffusion model, style textures are generated under the guidance of a single image, style information is extracted from a reference image, and meanwhile content information is ignored. The method comprises the following steps: firstly, obtaining a grid model of a three-dimensional object and a stylized image for guiding, obtaining text embedding characteristics and style characteristics, and completing style injection in a diffusion model cross attention mechanism in a layering manner; and optimizing texture parameters of the texture representation model by utilizing interval score matching loss and style guide loss, and finally, sampling the texture representation model to obtain the final stylized texture of the high-quality three-dimensional object.

A three-dimensional temperature gradient field visualization method comprises the following steps: after analyzing a part model and storing a point-surface relationship and temperature information, calculating a data point temperature gradient and carrying out point cloud, grid and entity model visualization; after creating temperature field color mapping and calculating a temperature field contour line, representing a temperature gradient field based on a point icon, representing a temperature gradient field based on a streamline and representing a temperature gradient field based on texture; through design of a data visualization core algorithm and model building, visualization of three-dimensional data of various vector fields for an unstructured grid model is realized.

Video data that depicts a face of a particular user is obtained. The video data is processed with a plurality of machine-learned models of a user-specific model ensemble for photorealistic facial representation to obtain a corresponding plurality of model outputs. The plurality of machine-learned models comprises one or more of a mesh representation model trained to generate a 3D polygonal mesh representation of the face of the particular user, a machine-learned texture representation model trained to generate a plurality of textures representative of the face of the particular user, or subsurface anatomical representation model(s) trained to generate sub-surface model outputs, each including a representation of a different sub-surface anatomy of the face of the particular user. At least one machine-learned model is optimized based on a loss function that evaluates the at least one model output.

The invention relates to the technical field of beef muscle texture recognition, and discloses a multi-modal image AI recognition method and system for beef muscle textures, and the method comprises the steps: collecting a beef image and spectral data; carrying out denoising, enhancement and normalization processing; using a variational auto-encoder to extract fusion features; optimizing a modal fusion strategy through reinforcement learning; reconstructing and enhancing a texture image by using the generative model; the key texture features are subjected to self-adaptive weighted reinforcement; the input classifier outputs a texture recognition result; the system comprises an image acquisition module, a spectrum acquisition module, a data preprocessing module, a multi-modal feature extraction module, a feature fusion module, an image generation module, a texture feature analysis module and an identification decision module. According to the invention, a bimodal joint modeling mechanism of image and spectral information is introduced, deep feature mapping is carried out on two types of modals by means of a variational auto-encoder, a cross-modal expression relationship is established through a submerged space, and the texture expression integrity and resolution are enhanced.

The application discloses a style image guided three-dimensional model texture generation method, a computer device, a storage medium and a program product. The style image guided three-dimensional model texture generation method is a pipeline based on a diffusion model, which generates a style texture under the guidance of a single image, extracts style information from a reference image, and ignores content information. First, a mesh model of a three-dimensional object and a style image used for guidance are obtained, text embedding features and style features are obtained, and style injection is completed in a diffusion model cross-attention mechanism in layers. The texture parameters of a texture representation model are optimized by using interval score matching loss and style guidance loss, and finally, the style texture of the high-quality three-dimensional object is obtained by sampling the texture representation model.

Texture filtering is applied to a texture represented with a mipmap comprising a plurality of levels, wherein each level of the mipmap comprises an image representing the texture at a respective level of detail. A texture filtering unit has minimum and maximum limits on an amount by which it can alter the level of detail when it filters texels from an image of a single level of the mipmap. The range of level of detail between the minimum and maximum limits defines an intrinsic region of the texture filtering unit. If it is determined that a received input level of detail is in an intrinsic region of the texture filtering unit, texels are read from a single mipmap level of the mipmap, and the read texels from the single mipmap level are filtered to determine a filtered texture value representing part of the texture at the input level of detail. If it is determined that the received input level of detail is in an extrinsic region of the texture filtering unit: texels are read from two mipmap levels of the mipmap, and the read texels from the two mipmap levels are processed to determine a filtered texture value representing part of the texture at the input level of detail.

The present application relates to the technical field of surface defect detection, and particularly relates to a surface detection method for shaft part machining, which screens a defect suspected area according to a cross-section diameter parameter obtained in each detection area along a characteristic direction, screens a risk profile according to a distribution coefficient of a texture profile, marks a first risk sub-profile or determines a characteristic risk profile based on a texture representation value of the risk profile, determines a second risk sub-profile according to a characteristic profile point and screens the characteristic risk profile, constructs a defect trend vector according to the risk profile to determine a profile trend representation coefficient, adjusts an illumination parameter for defect detection of the characteristic risk profile, and performs defect detection on the characteristic risk profile again based on the adjusted illumination parameter. The present application realizes rapid identification of a region with a risk of shadow obstruction, adaptively adjusts a detection mode according to actual defect characteristics of shaft parts, and improves the precision and reliability of surface defect detection.

The present invention belongs to the field of gait analysis and provides a method and system for identifying abnormal gait in a skeleton based on a generalized convolutional neural network. The method comprises constructing a gait graph using lower limb joints as nodes and skeletons as edges; grouping adjacent nodes in the gait graph based on distance and spatial relationships, and normalizing the non-Euclidean gait graph into a Euclidean feature graph, wherein the positional relationships of the nodes encode the connectivity between the joints; using a generalized convolutional network to model the geometric features encoded in the normalized Euclidean feature graph, generalizing the convolution kernel's ability to extract texture representations from images to extract geometric features from the gait graph, and obtaining a gait feature graph that aggregates the spatiotemporal neighbor node features; and predicting abnormal gait classification results based on the gait feature graph.

The application discloses a single-view three-dimensional texture consistency generation method based on geometric structure, pre-processes single-view image input data of a to-be-processed object and corresponding view angle parameters to obtain training input images, a view angle parameter set and supervision information, carries out geometric structure feature extraction and weight mapping processing on the training input images to obtain a geometric structure weight map, carries out sampling density distribution and position sampling on three-dimensional space texture query positions according to the geometric structure weight map to obtain a sampling point set, carries out multi-resolution coding on each sampling point in the sampling point set, and constructs a continuous texture representation through a multi-resolution texture hash grid to obtain texture attributes corresponding to each sampling point, and then carries out differentiable rendering, loss calculation and parameter updating to obtain a three-dimensional texture generation result. Compared with the prior art, the application improves the structure consistency, detail integrity and stability of three-dimensional texture generation under single-view conditions.

To provide a subband-based film grain evaluation process that determines an accurate evaluation score.SOLUTION: A method includes receiving a first image and a second image for film grain comparison, analyzing the first image or the second image to determine a first texture representation for the first image or a second texture representation for the second image, analyzing a set of regions on the basis of the first texture representation or the second texture representation, transforming the set of regions from the spatial domain to the frequency domain to generate a first frequency domain representation for the set of regions in the first image and a second frequency domain representation for the set of regions in the second image, and generating a score for an assessment of the difference in film grain in the first image and the second image on the basis of the first frequency domain representation and the second frequency domain representation.SELECTED DRAWING: Figure 9

A three-dimensional model processing method and apparatus, and a computer device, a computer-readable storage medium and a computer program product. The method comprises: acquiring a mesh representation of a three-dimensional model; on the basis of the mesh representation, performing model pre-processing on a first texture optimization model, so as to obtain a second texture optimization model; acquiring view parameters under a plurality of camera views in a world coordinate system, and on the basis of the mesh representation and the view parameters under the plurality of camera views, performing multi-layer texture optimization processing on the second texture optimization model, so as to obtain a third texture optimization model; using the third texture optimization model to perform texture prediction processing on the mesh representation, so as to obtain a target texture representation of the three-dimensional model; and exporting the target texture representation as a texture map resource of the three-dimensional model.

The invention relates to the technical field of hazardous chemical substance tank detection, in particular to a vehicle tank surface texture analysis method and system based on deep learning. The method comprises the following steps: S1, preprocessing a surface image or a video frame of a tank body of the hazardous chemical substance transport vehicle; s2, performing tank body region segmentation on the preprocessed image to obtain a tank body surface effective analysis region ROI; s3, obtaining a texture representation image of illumination decoupling; s4, obtaining texture features of view angle decoupling; s5, outputting a degradation thermodynamic diagram and a degradation index; s6, calculating a repair suspicion thermodynamic diagram and a repair suspicion index; and S7, outputting a tank body surface texture analysis result based on the degradation index and the repair suspicion index. According to the method, the stability analysis of the surface texture of the tank body under the complex environment condition is realized, the consistency of analysis results is ensured, and meanwhile, the surface state change caused by natural degradation and human intervention is effectively distinguished.

In some embodiments, a method receives a first image and a second image for a comparison of film grain. The first image or the second image is analyzed to determine a first texture representation of the first image or a second texture representation of the second image. A set of regions is analyzed based on the first texture representation or the second texture representation. The method converts the set of regions from a spatial domain to a frequency domain to generate a first frequency domain representation for the set of regions in the first image and a second frequency domain representation for the set of regions in the second image. A score is generated for an assessment of differences of the film grain in the first image and the second image based on the first frequency domain representation and the second frequency domain representation.

This invention provides an image semantic transmission method and system for remote sensing small target detection. The invention constructs an end-to-end semantic transmission detection model comprising a multi-scale feature extraction backbone network, a JSCC feature transmission network, a feature fusion module, and a detector. This model is deployed on an unmanned aerial vehicle (UAV) platform, and the pre-trained end-to-end semantic transmission detection model is used to detect transmitted remote sensing images and output target detection results. This invention employs feature-level JSCC continuous symbol transmission, avoiding hard decision failures in discrete links at low SNR, thus significantly mitigating the cliff effect. Gradient-based channel importance prioritizes the retention of high-frequency and texture representations most critical for small target detection under a fixed CBR, reducing the bandwidth / power consumption of redundant background channels. Policy distillation transfers privileged gradient knowledge obtained during training to a lightweight statistical predictor, enabling the generation of high-quality masks at the edge even without ground truth values.

The present invention is titled "Water Body Coding for a Vision System". The present invention provides a technically improved vision system and method for providing enhanced water coding. The vision system receives terrain data from a terrain database and aircraft state data from a navigation system. The vision system commands a display system to display a map image reflecting a view from the current position of the platform. The vision system processes the terrain data and the aircraft state data to identify water body features, determine the distance between the platform and the water body features, and determine a texture representation of the water body features. The texture representation includes symbols and associated symbol patterns. The vision system commands the display system to overlay the water body features on the map image with the texture representation.

The invention discloses an aerial photography small target detection method based on multi-scale context feature extraction, and relates to the technical field of image processing. The method comprises the following steps: acquiring aerial image data of an unmanned aerial vehicle and preprocessing the aerial image data; inputting the preprocessed image data into a backbone network for preliminary extraction to obtain basic features; the long-distance sensing module performs context enhancement processing on the basic features to obtain long-distance context features; a multi-scale sensing unit performs multi-scale semantic modeling on the long-distance context features to obtain multi-scale semantic features; the downsampling module performs light downsampling and feature compression on the multi-scale semantic features to obtain compact features; inputting the compact features into a feature fusion network for cross-scale information interaction to obtain fusion features; and inputting the fusion features into a detection head for classification and positioning prediction to obtain a detection result. According to the invention, through adaptive modeling of the multi-scale effective receptive field and texture expression of fine granularity, the detection precision and robustness of the aerial photography small target are significantly improved.

The invention discloses a granite texture surface layer pattern analysis method based on fractal dimension, and relates to the technical field of image recognition, and the method comprises the steps: obtaining texture images of a granite surface at a plurality of frequency bands, carrying out the multi-modal fusion, and generating an enhanced texture representation image; performing modal decomposition on the enhanced texture representation image to obtain a modal image sequence; the fractal dimension of each modal image in the modal image sequence is calculated, and a fractal heterogeneous feature set of granite texture is constructed; identifying a granite surface texture critical region based on the fractal heterogeneous feature set, and extracting a mineral crystal form feature set of the critical region; and constructing a comprehensive evaluation model based on the fractal isomerism feature set and the mineral crystal morphology feature set, predicting granite performance indexes and generating an analysis result. According to the method, a multi-band imaging and tensor decomposition combined feature fusion method is adopted, effective representation and fusion of cross-band features are realized through information entropy weight, and the integrity of the enhanced texture representation image is improved.

The application discloses a mesh cloth texture characterization method and system based on spatial frequency domain analysis, and belongs to the field of building material detection and image analysis. Stable texture base maps are obtained through illumination normalization and yarn boundary enhancement processing of mesh cloth texture maps. On the basis, multi-directional frequency scanning is carried out, frequency response distribution spectrum is constructed, and coherent peak trajectories are extracted. Further, based on the continuity and split features of the coherent peak trajectories, coherent break sections are identified, and yarn offset paths are back calculated. Through local expansion of the yarn offset paths, phase shift sequences are established, and combined with the frequency response distribution, energy closed-loop deviation degree indexes are constructed. Finally, through spatial continuity analysis, abnormal areas are identified, and texture distortion grades and structure weakening positions are output. The application can realize effective identification and positioning of subtle instability of the periodic structure of mesh cloth, and improve the accuracy and stability of texture defect detection.

The present application relates to a kind of lightweight low-light image enhancement methods based on illumination iterative adjustment, HVI color space transformation is carried out to the image to be processed to obtain HVI image, separates out illumination intensity component to obtain illumination intensity feature map, and shallow feature map is obtained by convolution processing to HVI image;Feature fusion and brightness adjustment are carried out to illumination intensity feature map and shallow feature map to obtain deep feature map, and size is recovered gradually by two layers of decoder stepwise up-sampling, and channel adjustment is carried out using specified size convolution in each layer of decoder, and up-sampling is realized using bilinear interpolation;Mamba-based structure refinement module is introduced at the end of each layer of decoder to improve structure restoration capability, and the HVI feature output by the last layer of decoder is adjusted to specified value after a specified size convolution is carried out, and residual connection is carried out with initial HVI image, to obtain enhanced HVI image, and recovery image is obtained after inverse HVI transformation.In each low-light condition, adaptive enhancement is carried out, multi-directional spatial context aggregation is realized, and texture representation is improved.

The invention discloses a multi-scale feature fusion semantic segmentation method and a multi-scale feature fusion semantic segmentation system based on a diffusion model and Transform. The method comprises the following steps: carrying out normalization and data enhancement preprocessing on an input image; extracting multi-scale detail features in a plurality of time steps through a pre-trained diffusion model, and forming a diffusion feature set through channel compression and normalization; a visual Transform module is utilized to extract global semantic features; the method comprises the following steps: by taking a Transform feature as a query and a diffusion feature as a key value (Key / Value), realizing semantic-guided detail fusion through a trans-attention mechanism; performing context enhancement in combination with an enhanced local attention (ELA) and scale awareness pyramid pooling (SFA-PPM) module, and constructing a multi-scale feature pyramid; and finally, outputting a pixel-level semantic segmentation result through the lightweight MLP decoder. According to the method, the fine-grained texture representation of the diffusion model and the global semantic modeling capability of Transform are effectively combined, and the segmentation precision, the boundary fidelity and the small target recognition performance are improved. The method is high in calculation efficiency, can realize real-time reasoning on the middle-end GPU, and is suitable for the fields of automatic driving, remote sensing analysis, medical image segmentation and the like.

This invention relates to the field of image processing technology, specifically to an artificial intelligence-based method and system for detecting the maturity of melons. The method acquires side images of the melon from multiple angles to obtain corresponding principal component side images. Two-stage wavelet transforms are performed on the principal component side images to obtain detail images in different directions. The gradient changes of the detail images in corresponding directions are used to obtain the corresponding step size, thereby obtaining the corresponding gray-level co-occurrence matrix. The feature parameters of the gray-level co-occurrence matrices in the three directions form the texture representation matrix of the principal component side images. The average saturation, average pixel value, pixel value variation range, and gray-level distribution uniformity of all pixels in the side images constitute a color feature matrix. The texture representation matrix and the color feature matrix form the feature analysis matrix for each side image. The maturity of the melon is detected based on the feature analysis matrix. This invention can accurately assess the maturity of melons with high recognition accuracy.

This invention discloses a multi-temporal super-resolution reconstruction method with spatial context enhancement, belonging to the field of remote sensing image processing technology. The method first acquires multi-temporal low-resolution remote sensing image sequences and performs radiometric calibration, atmospheric correction, geometric correction, and spatial registration to construct a standardized temporal dataset. Then, a semantic prior extraction module is constructed to generate a scene semantic distribution map. Further, differentiated feature enhancement and multi-temporal fusion are performed on different land cover regions. Constraints on spectral consistency, spatial structure preservation, semantic consistency, and texture realism are introduced to establish a joint optimization framework for super-resolution reconstruction and semantic parsing tasks. Finally, the trained model is used to output high-resolution reconstruction results and a semantic map. This invention can effectively improve the structural recovery capability and texture representation capability of different land cover regions, and has significant application value in resource surveys, land cover identification, disaster monitoring, and other fields.

The present application discloses a method, apparatus, device, and storage medium for extracting feature point descriptors for weak texture representation. The method comprises: inputting a liver image of a preset size; processing the liver image through a Gaussian filter unit and a filter unit to generate a feature descriptor; and outputting the feature descriptor and the corresponding corner point coordinates. The above-mentioned solution provided by the present application can extract more effective feature point pairs and, based on additional information such as the extracted angle, size, and distance, add a filter layer to effectively filter out invalid feature point pairs. At the same time, the present invention adopts accelerated computing on a PYNQ board, and the result can reach 60 frames / s, meeting the real-time requirements of the system.

The invention relates to the technical field of image segmentation, in particular to a bridge surface disease detection system based on image semantic segmentation, and the system comprises a bridge surface texture quantification module which is used for setting the size of a sliding window associated with the particle size of concrete aggregate, constructing pixel gray values into a three-dimensional surface in each sliding window, and outputting the three-dimensional surface; and counting the number of boxes of different sizes covering the three-dimensional surface, calculating the relationship slope of the number of the boxes and the sizes of the boxes under logarithmic coordinates, and establishing a fractal dimension value of the local surface. According to the method, on the basis of image semantic segmentation, sliding windows corresponding to the particle size of concrete aggregate are established, pixel gray values are mapped into three-dimensional surfaces in each sliding window, and the logarithmic slope between the number of boxes and the size is extracted through multi-scale box coverage statistics, so that the microcosmic difference of the surface texture is expressed; a texture representation graph is obtained after fractal dimension distribution is summarized in the whole image, and the perception capability of local structure abnormity is enhanced.

The invention relates to a printing technology, and discloses a medium type self-adaptive printing method based on multispectral recognition, which comprises the following steps of: acquiring a multispectral medium image of a medium to be printed, calculating a binary characteristic value corresponding to the multispectral medium image, extracting medium texture representation in the multispectral medium image according to the binary characteristic value, and printing the medium texture representation according to the extracted medium texture representation. Performing block processing on the multispectral medium image to obtain a block medium image; geometric configuration features corresponding to the block medium images are identified, and medium contour features corresponding to the block medium images are analyzed; wave band reflectivity values of the partitioned medium images are calculated to construct medium reflection curves corresponding to the partitioned medium images, and after spectral absorption characteristics corresponding to the partitioned medium images are analyzed based on the medium reflection curves, the medium type corresponding to the to-be-printed medium is determined; and in combination with the medium contour feature and the medium type, printing processing on the to-be-printed medium is executed, and a printing result is obtained. The invention aims to improve the printing effect of the printing medium.

A method of visualizing treatment of a denim fabric is provided. The method comprises, at an electronic device with a display, retrieving a source texture, the source texture representing a pixel map of a scan of the denim fabric, displaying a user interface on the display, the user interface comprising a set of user interactive elements for adjusting values of parameters of the source texture, wherein each user interactive element corresponds to one of the parameters of the source texture, displaying, in the user interface, a 3D model having a surface comprising a representation of the source texture, receiving one or more user inputs corresponding to one or more interactions with at least one user interactive element of the set of user interactive elements, the one or more interactions comprising adjustment of values of at least one of the parameters of the source texture to obtain adjusted values of the at least one of the parameters of the source texture, in response to receiving the one or more user inputs, processing the source texture based on the adjusted values of the at least one of the parameters to obtain an adjusted source texture and displaying, in the user interface, the 3D model wherein the surface comprises a representation of the adjusted source texture. An electronic device and a computer-readable storage medium are also provided.

The invention discloses a semantic communication-oriented high texture fidelity image compression method and system independent of a sending end model, and the method comprises the steps: carrying out the extraction of semantic information, structural information, an entity region and texture information for a given original image, and obtaining a semantic representation, a structural representation, an entity mask set and a texture representation; based on the semantic representation, the structure representation, the entity mask set and the texture representation, performing rate-distortion optimization on the structure representation to obtain an optimal structure residual error, and then encoding the semantic representation, the texture representation and the optimal structure residual error to obtain a code stream; and at a decoding end, decoding the code stream to obtain a semantic representation, a lossy structure residual error and a lossy texture representation, and then reconstructing an image through a multi-modal image generation model. According to the method, the dependence of encoding and decoding ends on model synchronization can be effectively reduced, the texture fidelity of a reconstructed image is improved, and the bit rate is reduced.