Micro-Block Matching Coding for Image Compression Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image coding and decoding technologies face inefficiencies due to low compression and decompression efficiency, particularly in handling similar patterns within computer screen images, leading to high complexity, large computational demands, and excessive memory bandwidth usage.
Innovation Solution
The method involves performing micro-block matching coding on Coding Units (CUs) to generate optimal Displacement Vectors (DVs), followed by individual and global transformations to reduce redundancy, and then integrating these into the standard coding and decoding processes, including intra and inter prediction, quantization, and entropy coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If intra block copy manner with fixed sizes (8×8, 16×16, 32×32 and 64×64 pixels) is used, then accurate matching of various sizes and shapes can be implemented, but device complexity and computational amount increase significantly
Solution Approach 1:
The patent divides the current block into multiple sub-blocks and performs matching on each sub-block independently. This segmentation allows the system to achieve accurate matching for various sizes and shapes by treating different regions separately, while keeping the complexity of each individual sub-block matching operation manageable through the use of fixed template sizes.
2Measurement precision
If siring matching manner is used to implement accurate matching of various sizes and shapes, then matching precision improves, but computational amount and memory read-write bandwidth increase greatly
Solution Approach 1:
The patent uses template copying where a fixed-size template is copied and compared against sub-blocks in the current block. This copying approach enables accurate matching of various sizes and shapes by adjusting the template position and size parameters, while maintaining reasonable computational requirements through the efficient reuse of the same template across multiple comparisons.
3Device complexity
If existing intra prediction manner is used, then device complexity is controlled, but compression efficiency is limited due to inability to utilize similarity in the same frame
Solution Approach 1:
The patent extends the prediction approach by searching for matching blocks within the same frame (intra-frame search) in addition to using adjacent pixel samples. This adds a spatial dimension to the prediction process, allowing the system to exploit self-similarity patterns within the frame while maintaining controlled complexity through the use of fixed template sizes and systematic search strategies.
Data Source
AI summary
The disclosure discloses a method and device for coding an image and a method and device for decoding an image. The coding includes that: a coding block is partitioned into L micro-blocks, wherein L is a preset number of micro-block partitions in the decoding block; Displacement Vectors (DVs) between the micro-blocks and reference micro-blocks in a preset reconstructed pixel sample set are determined, wherein the DVs are vectors between pixel positions of the reference micro-blocks and pixel positions of the micro-blocks in a preset pixel coordinate system: and the DVs are coded and written into a bitstream. By the methods and the devices, the problem of low image compression and decompression efficiency during image coding and decoding in the related technology is solved, and tire effect of improving image coding and decoding efficiency is further achieved.