Image Coding with Region-Specific Probability Updates
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 utilizing spatially close blocks for more accurate probability calculations.
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, but parallel processing capability deteriorates
Solution Approach 1:
The image is divided into multiple slices, and each slice is independently coded with its own probability table. This segmentation allows different slices to be processed in parallel while maintaining coding efficiency within each slice through localized probability updates.
Solution Approach 2:
Each slice maintains its own probability table that is updated locally based on the coding progress within that slice. This local quality approach ensures that probability information remains relevant to the current coding context while enabling independent parallel processing of multiple slices.
2Productivity
If slices are processed independently without spatial correlation, then parallel processing is enabled, but coding efficiency deteriorates
Solution Approach 1:
Probability tables are initialized before slice processing begins, with preliminary values that can be updated during coding. This preliminary action setup enables parallel processing while maintaining the capability for efficient probability updates within each slice context.
3Measurement precision
If probability tables are updated continuously in raster order, then accurate probability information is maintained, but operation frequency and processing speed are limited
Solution Approach 1:
The continuous probability update process is segmented into discrete slice-based updates. Each slice processes its own probability table independently, allowing higher operation frequencies within slices while maintaining overall accuracy through localized updates.
Solution Approach 2:
Probability table updates occur periodically at slice boundaries rather than continuously across the entire image. This periodic action within slices maintains probability accuracy while enabling higher processing speeds through parallel slice execution.
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.