Image Coding Probability Initialization Across Parallel Slices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image coding methods, such as the H.264 standard, face challenges in improving coding efficiency due to the need to code macroblocks in raster order while updating probability tables, which limits parallel processing and operation frequency, and fails to utilize spatial correlation between slices effectively.
Innovation Solution
An image coding method that updates probability information based on the characteristics of each region within an image, allowing for sequential coding of blocks using region-specific probability information, which is updated after coding each block to improve coding efficiency by using more appropriate probability data for adjacent blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If macroblocks are coded in raster order with continuous probability table updates, then coding efficiency is improved through adaptive probability modeling, but parallel processing capability deteriorates and operation frequency is limited
Solution Approach 1:
The image is divided into multiple slices that can be processed in parallel. Each slice maintains its own probability table that is initialized based on spatially adjacent slices, enabling independent parallel processing while preserving adaptive probability modeling benefits through inter-slice probability sharing.
Solution Approach 2:
Probability tables for subsequent slices are pre-initialized using probability information from spatially adjacent slices before processing begins. This preliminary action allows parallel processing to start immediately with informed probability models, eliminating the need for sequential probability table updates.
2Productivity
If slices are processed independently to enable parallel decoding, then processing speed is improved, but spatial correlation between slices is not utilized and coding efficiency deteriorates
Solution Approach 1:
Probability information from spatially adjacent slices is fed back as initialization data for subsequent slice processing. This feedback mechanism allows independent parallel slice processing while incorporating spatial correlation information to maintain high coding efficiency.
Solution Approach 2:
Probability tables are pre-initialized using information from spatially adjacent slices before independent parallel processing begins, allowing slices to be processed independently while still utilizing spatial correlation for improved coding efficiency.
3Ease of operation
If macroblocks are coded in raster order, then processing simplicity is maintained, but spatial correlation between distant macroblocks is not utilized and coding efficiency deteriorates
Solution Approach 1:
Probability tables are pre-initialized using information from spatially adjacent macroblocks or slices before processing begins. This preliminary action enables the system to utilize spatial correlation without requiring complex sequential processing, maintaining relative simplicity while improving coding efficiency.
Data Source
AI summary
An image coding method for improving coding efficiency by using more appropriate probability information is provided. The image coding method includes: a first coding step of coding a first set of blocks included in a first region sequentially based on first probability information; and a second coding step of coding a second set of blocks included in a second region sequentially based on second probability information. In the first coding step, the first probability information is updated depending on data of a target block to be coded, after coding the target block and before coding a next target block. In the second coding step, the second probability information is updated depending on the first probability information updated in the first coding step, before coding the first target block.