CN-121810824-B - Dynamic texture atlas reorganization compression method and system based on airspace sparsity

CN121810824BCN 121810824 BCN121810824 BCN 121810824BCN-121810824-B

Abstract

The invention discloses a dynamic texture atlas reorganization compression method and a system based on airspace sparsity, which belong to the technical field of data compression, and comprise the steps of obtaining a dynamic texture sequence, carrying out feature analysis on texture units in the dynamic texture sequence, and generating multidimensional feature vectors; the method comprises the steps of constructing an initial compressibility potential model for each texture unit based on multidimensional feature vectors, executing iterative collaborative decision process by taking the initial compressibility potential model as input to generate optimized texture atlas layout and a partition compression strategy mapping table, executing differential compression on texture data arranged according to the optimized texture atlas layout according to the partition compression strategy mapping table to generate compressed texture data, integrating and packaging layout information of the optimized texture atlas layout, the partition compression strategy mapping table and the compressed texture data to generate a compressed data stream. The invention adopts a collaborative decision mechanism and combines airspace sparsity analysis and a differential compression strategy, thereby realizing the efficient compression of the dynamic texture atlas.

Inventors

LING YUN
XU MING
ZHANG XIAO
Gou Zhun
WANG CHUAN
CUI DAGUANG
LIN MAI
JIANG ZHENXIU

Assignees

成都力比科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260312

Claims (7)

1. The dynamic texture atlas reorganization compression method based on airspace sparsity is characterized by comprising the following steps: Obtaining a dynamic texture sequence, and carrying out feature analysis on texture units in the dynamic texture sequence to generate respective multidimensional feature vectors, wherein the method comprises the steps of calculating pixel distribution complexity in the texture units to generate consistency features in a space domain, quantifying relevance of the texture units and other units in a space-time neighborhood of the texture units to generate neighborhood relevance features, and fusing the consistency features in the space domain and the neighborhood relevance features into respective multidimensional feature vectors; Constructing an initial compressibility potential model for each texture unit based on the multidimensional feature vector, wherein the initial compressibility potential model is used for representing the compression potential of the texture unit in different neighborhood environments; The method comprises the steps of taking an initial compressibility potential model as input, executing an iterative collaborative decision process, synchronously generating an optimized texture atlas layout and a partition compression strategy mapping table matched with the optimized texture atlas layout in the process, generating an initial atlas layout draft according to the initial compressibility potential model, executing a joint decision cycle on the initial atlas layout draft, generating a predicted compression result through simulation compression, calculating a joint efficiency score based on the predicted compression result, adjusting the initial atlas layout draft according to the joint efficiency score, generating an updated atlas layout draft, repeatedly executing the joint decision cycle and the adjustment step until the joint efficiency score meets a preset convergence condition, and taking the updated atlas layout draft and a corresponding compression strategy as the optimized texture atlas layout and the partition compression strategy mapping table; Performing differential compression on the texture data arranged according to the optimized texture atlas layout according to the partition compression strategy mapping table to generate compressed texture data, wherein the differential compression comprises the steps of analyzing the partition compression strategy mapping table, identifying a first area and a second area in the optimized texture atlas layout, wherein the first area corresponds to a first compression algorithm, and the second area corresponds to a second compression algorithm; and integrating and packaging the layout information of the optimized texture atlas layout, the partition compression strategy mapping table and the compressed texture data to generate a compressed data stream.
2. The spatial sparsity-based dynamic texture atlas rebinning compression method of claim 1, wherein the constructing the initial compressibility potential model comprises: Obtaining a group of candidate neighborhood environment templates representing typical neighborhood environments; Combining the multidimensional feature vector of the texture unit with each of the candidate neighborhood environment templates to form a set of combined feature vectors; Based on the group of combined feature vectors, potential compression gain coefficients corresponding to each combined feature vector are obtained; and mapping and correlating the candidate neighborhood environment templates with the corresponding potential compression gain coefficients to form an initial compressibility potential model.
3. The dynamic texture atlas reorganization compression method based on airspace sparsity of claim 1, wherein the calculating a joint efficiency score includes: calculating the data compression degree from the estimated compression result to generate a compression rate factor; Evaluating the fidelity difference between the estimated compression result and texture data to generate a quality loss factor; simulating a texture sampling process of the initial atlas layout draft, evaluating cache access efficiency of the initial atlas layout draft, and generating a sampling efficiency factor; and carrying out weighted combination on the compression rate factor, the quality loss factor and the sampling efficiency factor according to a preset weight coefficient to obtain a joint efficiency score.
4. The spatial sparsity-based dynamic texture atlas reorganization compression method of claim 1, wherein the applying the first compression algorithm to the texture data in the first region includes: Calculating the average value of all pixels in the high-homogeneity sparse cluster to obtain a representative basic value; calculating the difference value between each pixel value in the high-homogeneity sparse cluster and the representative basic value to generate a differential matrix; and respectively carrying out entropy coding on the representative basic value and the differential matrix to generate compressed texture data of a first area.
5. The spatial sparsity-based dynamic texture atlas reorganization compression method of claim 1, wherein the generating a compressed data stream includes: converting the layout information of the optimized texture atlas layout into a layout index table; Converting the partition compression strategy mapping table into a strategy instruction sequence; creating a metadata block, and storing the layout index table and the strategy instruction sequence into the metadata block; and splicing the metadata block and the compressed texture data to form a compressed data stream.
6. The spatial sparsity-based dynamic texture atlas reorganization compression method of claim 5, further comprising: analyzing the compressed data stream, and separating the metadata block and the compressed texture data; according to the strategy instruction sequence in the metadata block, the compressed texture data are decompressed in parallel, and an optimized texture atlas is reconstructed; and when rendering and sampling, converting the logical sampling coordinates from the original texture sequence into physical sampling coordinates on the optimized texture atlas in real time by utilizing a layout index table in the metadata block.
7. A dynamic texture atlas reorganization compression system based on airspace sparsity, which is applied to a dynamic texture atlas reorganization compression method based on airspace sparsity as claimed in any one of claims 1 to 6, and is characterized in that the system comprises: the texture feature analysis module is used for acquiring a dynamic texture sequence, carrying out feature analysis on texture units in the dynamic texture sequence and generating respective multidimensional feature vectors, wherein the texture feature analysis module comprises the steps of calculating pixel distribution complexity in the texture units and generating consistency features in a space domain; The potential model modeling module is used for constructing an initial compressibility potential model for each texture unit based on the multidimensional feature vector, wherein the initial compressibility potential model is used for representing the compression potential of the texture unit in different neighborhood environments; The collaborative decision-making and layout optimization module is used for taking the initial compressibility potential model as input, executing an iterative collaborative decision-making process, synchronously generating an optimized texture atlas layout and a partition compression strategy mapping table matched with the optimized texture atlas layout in the process, wherein the collaborative decision-making and layout optimization module comprises the steps of generating an initial atlas layout draft according to the initial compressibility potential model, executing a joint decision-making cycle on the initial atlas layout draft, generating a predicted compression result through simulation compression, and calculating a joint efficiency score based on the predicted compression result; The differential compression module is used for performing differential compression on the texture data arranged according to the optimized texture atlas layout according to the partition compression strategy mapping table to generate compressed texture data, and comprises the steps of analyzing the partition compression strategy mapping table, identifying a first area and a second area in the optimized texture atlas layout, wherein the first area corresponds to a first compression algorithm, and the second area corresponds to a second compression algorithm; And the data packaging module is used for integrating and packaging the layout information of the optimized texture atlas layout, the partition compression strategy mapping table and the compressed texture data to generate a compressed data stream.

Description

Dynamic texture atlas reorganization compression method and system based on airspace sparsity Technical Field The invention relates to the technical field of data compression, in particular to a dynamic texture atlas reorganization compression method and system based on airspace sparsity. Background In modern computer graphics and real-time rendering applications, dynamic textures play a vital role, which can bring a vivid visual effect to a scene, such as flowing liquid, a waving flag, or an animated character's expression. To manage and efficiently utilize these texture resources, a developer typically organizes multiple independent texture sequence units into one or more large texture atlases. Because the dynamic texture atlas contains pixel data which changes along with time, a large amount of storage space and memory bandwidth are occupied, and efficient compression of the dynamic texture atlas is a key link for improving application performance. In the related art, china patent publication No. CN120047590B discloses a self-adaptive texture atlas compression system and method based on a Unity engine, wherein the self-adaptive texture atlas compression system comprises a texture resource import module, a texture resource setting module, an atlas construction module and an atlas construction module, wherein the texture resource import module converts PNG into DYN files and generates configuration data, the texture resource setting module adopts a double-buffer configuration management mechanism, the improved CityHash algorithm is used for detecting hash value differences and asynchronously and synchronously changing fields, the atlas construction module is used for performing edge detection, polygon generation and rasterization processing, the active edge table algorithm is used for outputting pixel data, an arrangement unit is used for realizing mutually exclusive arrangement of the pixel data based on multithreading parallel computation and outputting a UV coordinate mapping table, the atlas setting module is used for dynamically modifying configuration parameters and triggering pipeline updating, geometric accuracy is ensured through vertex number and texture area verification, expansion offset and step length parameters are dynamically adjusted to adapt to different resolutions, and the integral breakpoint continuous transmission mechanism is used for guaranteeing the integrity of a transaction log. However, the above prior art scheme has the following technical drawbacks. The feature analysis dimension is single, the comprehensive compression potential of the texture unit is not fully mined, the prior art only focuses on the basic attribute or single dimension feature of texture resources, and the multidimensional features of the texture unit on the space domain consistency and the neighborhood correlation are not analyzed by the system. The method cannot accurately represent the differences of different texture units in dynamic change, internal pixel distribution and neighborhood association, so that the follow-up compression strategy lacks pertinence, texture clusters with high compression potential are difficult to identify, and compression gain cannot be maximized. The layout and the compression strategy lack of cooperative optimization and are easy to fall into local optimization, wherein the prior art splits the atlas layout construction and the compression algorithm selection into independent flows, the layout design is only focused on the compactness of pixel arrangement, a feedback mechanism of a compression effect is not introduced, and the compression algorithm selection is not dynamically adjusted according to the layout characteristics. The split decision results in the failure to form a globally optimal 'layout-compression' combination, which may result in the problems that the layout satisfies the arrangement efficiency but has high compression redundancy, or the compression algorithm has poor suitability but the layout limits the optimization space, and the overall compression efficiency is limited. The compression algorithm is selected and solidified, and differential precise compression is not realized, wherein the prior art adopts a uniform or limited compression algorithm to process the whole texture atlas, and the differential algorithm is not adapted to the content characteristics of different areas in the texture atlas. For a sparse region with high compression potential, extremely compression cannot be realized by utilizing the spatial sparsity of the sparse region, and for a complex detail region, the general algorithm easily causes excessive mass loss or insufficient compression rate, so that the compression efficiency and the visual fidelity are difficult to balance. Buffer access efficiency and decompression rendering compatibility are not fully considered, namely buffer access logic in the process of sampling the non-simulated texture is not designed according to the lay