13 results about "Compression artifact" patented technology

The invention provides an image compression artifact removal method based on a frequency domain hybrid expert network, and relates to the field of image processing. The method comprises the following steps: acquiring image data with compression artifacts; extracting shallow image features of the image data through an independent convolutional layer; extracting long-distance dependent image features of the shallow image features through a spatial attention mechanism; based on long-distance dependent image features, inputting the long-distance dependent image features into a frequency domain transformation expert network, and determining an optimal frequency domain transformation expert through parallel processing of multiple frequency domain transformation experts and an expert selection mechanism; processing the long-distance dependent image features through an optimal frequency domain transformation expert to obtain enhanced image features; performing image reconstruction on the enhanced image features to obtain a reconstructed image; the reconstructed image is an image after the compression artifacts are removed. The method is used in an image compression artifact removal process, and solves the technical problem that multiple types of complex compression artifacts are difficult to effectively remove in the prior art.

The invention discloses an optimization method and system for compressed point cloud data, equipment and a medium. The method comprises the following steps: acquiring first data and second data; extracting local geometric features and global structural features from the first data; adapting a compression distortion type of the first data according to the coding parameter; based on the local geometric features, determining a deviation range of each discrete point in the first data and a point missing region in the first data; restoring the first data by compressing a restoration network corresponding to the distortion type based on the global structure feature, the deviation range and a point missing region in the first data to obtain second data; the first data and the second data are merged to obtain optimized data, distortion types under different compression strengths can be adapted through the coding parameters, and targeted restoration and optimization of the compressed point cloud data are realized in combination with local and global features, so that the geometric fidelity and visual quality of the compressed point cloud data are improved.

The application provides a compression artifact suppression method combined with multi-level interframe correlation. The method mainly relates to a means combining multi-level motion compensation and deep convolution mapping to perform quality enhancement on HEVC decoded video. The application mainly utilizes the characteristics of interframe correlation of the video, and then designs a multi-level motion compensation network to perform motion compensation on two adjacent frames of a current frame, and then fuses features of the current frame with the two adjacent frames. Then, attention feature extraction is performed on the current frame and the fused multi-frame feature map respectively. Finally, deep residual learning and cross-layer cascade are performed to realize deep fusion, mapping and reconstruction of multi-level features, so that quality enhancement of the current frame is realized.

Systems and techniques are generally described for selecting a machine learning model for compression artifact removal and resolution upscaling of video streaming data. In various examples, a system or method receives a stream of video data, determines a category of the stream of video data based at least partially upon a compression level of the stream of video data, selects weights for a machine learning model based upon the category, and executes the machine learning model with the selected weights to remove compression artifacts in the stream of video data and upscale a resolution of the stream of video data.

A system and method for complex-valued radar image compression integrates AI-based techniques to enhance compression quality. It incorporates a novel AI deblocking network composed of convolutional layers for feature extraction and a channel-wise transformer with attention to capture complex inter-channel dependencies. The convolutional layers extract multi-dimensional features from the complex-valued radar image, while the channel-wise transformer learns global inter-channel relationships. This hybrid approach addresses both local and global features, mitigating compression artifacts and improving image quality. The model's outputs enable effective complex-valued radar image reconstruction, achieving advanced compression while preserving crucial information for accurate analysis.

Systems and methods are provided herein for efficient, reproducible noise sampling coordinate generation for texture synthesis in image and video processing. The techniques enable high-quality film grain or texture synthesis as a post-processing step in video decoders, masking compression artifacts and restoring perceptual quality. Hash-based mapping, including direct coordinate mapping and linear index hashing, can be used to generate coordinates for noise sampling by hashing image block coordinates and a seed with preselected constants to generate sampling positions. Shuffled permutation sampling can be used to generate coordinates by generating and shuffling a list of template coordinates using a seeded pseudo-random number generator, and assigning unique positions to each block. Wrap-around addressing, including treating the noise template as a toroidal space, can be used to maximize coverage and eliminate edge bias. The disclosed techniques provide deterministic, hardware-friendly, and unbiased sampling.

The invention discloses a data set expansion method, device and system, and a storage medium. The method comprises the following steps: S1, obtaining an original HSI-RGB data set; s2, performing spectrum enhancement on the original HSI-RGB data set to obtain an enhanced HSI-RGB data set; wherein the spectrum enhancement comprises synchronous space geometric enhancement, reverse optical rendering of physical perception, cross-spectral domain generative enhancement and real noise and compression artifact simulation. By adopting the technical scheme of the invention, a large number of HSI-RGB pairing data sets which are physically consistent and diversified and are accurately aligned in space can be generated from limited HS truth values, so that the generalization ability and robustness of a spectrum reconstruction model based on deep learning in a real world scene are improved.

The invention discloses a super-division method and system for reconstructing a high-quality 4K video through low-quality description, and relates to the technical field of video image processing, and the method comprises the steps: obtaining continuous frames of a to-be-processed video, constructing a time sequence frame set, carrying out the extraction of a compression block feature, a ringing feature and a noise feature of a target frame, and generating a spatial degradation map; recognizing a continuous degradation region according to the spatial degradation map and generating a spatial weight map; inputting the time sequence frame set into a space-time reconstruction network, performing structure reconstruction to obtain a basic reconstruction image, and generating a high-frequency residual image; and performing spatial weighting adjustment on the high-frequency residual image according to the spatial weight map, calculating a global degradation index of a target frame according to the spatial degradation map to determine a time sequence consistency threshold, constraining residual change between adjacent frames, and synthesizing the high-frequency residual image and the basic reconstruction image to obtain a 4K super-resolution video frame. Compression artifacts are suppressed and video inter-frame flicker is reduced while 4K video super-division reconstruction is completed.

A system and methods for upsampling of decompressed data after lossy compression using a neural network that integrates AI-based techniques to enhance compression quality. It incorporates a novel AI deblocking network composed of convolutional layers for feature extraction and a channel-wise transformer with attention to capture complex inter-channel dependencies. The convolutional layers extract multi-dimensional features from the two or more correlated datasets, while the channel-wise transformer learns global inter-channel relationships. This hybrid approach addresses both local and global features, mitigating compression artifacts and improving decompressed data quality. The model's outputs enable effective data reconstruction, achieving advanced compression while preserving crucial information for accurate analysis.

The present application provides a kind of lightweight super-resolution reconstruction model and system for compressing image, and the lightweight super-resolution reconstruction model includes: the feature representation of HR image is trained with C-LR image in compression artifact removal submodule;Wherein, the feature representation of HR image is extracted from HR image using VGG pre-training model;Image super-resolution submodule recovers the LR image that compression artifact removal submodule is output as input, and carries out image super-resolution by residual information distillation network.The present application constructs lightweight super-resolution reconstruction network, designs objective function and loss function, adds feature representation to improve the details of reconstructed image, improves the quality of reconstructed image, and finally high-definition high-resolution image.

The application provides a cross-domain contrast learning network for image tampering positioning. The network is composed of an image domain branch, a frequency domain branch, a feature fusion module, an attention enhancement module, and a cross-domain contrast loss module. The image domain branch takes ViT as the backbone network to model the spatial position relationship between pixels; the frequency domain branch describes compression artifacts based on discrete cosine transform (DCT) to extract more subtle tampering traces. The dual-domain features are fused to form fusion domain features, which are input into the channel-spatial multi-scale attention module enhanced by the median to capture local details and global context information simultaneously. Then, the enhanced multi-domain features are sent to the cross-domain contrast learning module to effectively distinguish the tampering area and the real area through intra-domain contrast learning, enhance the aggregation of positive samples, and separate the distance of negative samples; at the same time, the feature representation consistency is realized through cross-domain pair alignment to reduce the distribution difference between domains.

An image processing apparatus includes a memory storing at least one instruction; and a processor configured to execute the at least one instruction to use at least one neural network to: extract n pieces of first feature information from a first image, based on locations of pixels included in the first image, wherein n is a positive integer, generate n pieces of second feature information by performing a convolution operation between each of the n pieces of the first feature information and each of n kernels, and generate, based on the n pieces of the second feature information, a second image from which compression artifacts included in the first image are removed.

The application provides an image diffusion enhancement method and device based on multi-scale feature extraction and fusion, which comprises the following steps: obtaining a to-be-processed image; performing multi-scale feature extraction on the to-be-processed image to obtain a multi-scale feature map; performing pyramid feature fusion on the multi-scale feature map to obtain a fused feature map; performing compression block detection on the to-be-processed image and generating a block boundary mask; based on a preset retrieval algorithm, retrieving a similar reference image of the to-be-processed image from a natural image reference library and extracting a reference feature of the similar reference image; taking the fused feature map, the block boundary mask and the reference feature as conditions, using a conditional latent space diffusion model to enhance the to-be-processed image and output an enhanced image. The method and device of the application can effectively restore the brightness, contrast and detail texture of a downhole low-light-intensity compressed image without relying on paired training data, significantly suppresses compression artifacts such as color blocks and ringing, and generates a high-quality enhanced image with rich details and natural structure.