Depth Map Coding via Block Segmentation for Simplified Depth C
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for coding and decoding depth maps in three-dimensional video images are limited by a maximum prediction size, which restricts the selection of intra-frame prediction modes and decreases coding efficiency when image blocks exceed this size, leading to increased design overheads and reduced processing efficiency.
Innovation Solution
The method involves splitting larger image blocks into smaller blocks of maximum prediction size for simplified depth coding (SDC), allowing the use of intra-frame prediction modes like depth modeling mode, planar mode, or direct current mode, and calculating average prediction data and residuals to facilitate efficient decoding and coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the maximum prediction size is limited to 32×32, then the coding system design is simplified, but the coding efficiency decreases when image blocks are larger than 32×32
Solution Approach 1:
The patent divides a large image block (e.g., 64×64) into multiple smaller prediction blocks (e.g., four 32×32 blocks) that fit within the maximum prediction size limit. This segmentation allows the coding system to process each sub-block using the supported maximum prediction size, thereby maintaining design simplicity while improving coding efficiency for large image blocks.
2Quantity of substance
If the image block size is greater than the maximum prediction size, then more data can be processed in each block, but the DMM cannot be selected resulting in decreased coding efficiency
Solution Approach 1:
The patent segments a large image block into multiple smaller prediction blocks that fit within the maximum prediction size constraint. This enables the Depth Modeling Mode (DMM) to be applied to each sub-block, thereby maintaining high coding efficiency while processing larger overall data volumes through multiple smaller blocks rather than a single large block.
3Adaptability or versatility
If image blocks larger than maximum prediction size are introduced, then the coding coverage is increased, but the design overheads of the coding and decoding system increase
Solution Approach 1:
The patent maintains the maximum prediction size limit (e.g., 32×32) to avoid increasing design overheads, while achieving increased coding coverage by processing large image blocks through multiple segmented prediction blocks. This approach preserves system simplicity while extending the effective processing capability to larger image dimensions.
Data Source
AI summary
Embodiments of the present invention provide methods and devices for coding and decoding a depth map, including: determining to perform simplified depth coding (SDC) decoding according to a flag of an SDC mode; determining a size of an image block and a maximum prediction size; determining an intra-frame prediction mode; in a case in which the size of the image block is greater than the maximum prediction size, splitting the image block to obtain N split image blocks; and performing the SDC decoding on the N split image blocks by using the intra-frame prediction mode. In this way, processing efficiency of coding and decoding a depth map can be improved.