Adaptive Image Coding via Segmented Quantization Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image coding technologies lack an effective method to adaptively code continuous-tone and discontinuous-tone content according to different reconstruction qualities and distortion degrees, resulting in either high bit rates due to maintaining high quality for continuous-tone content or significant distortion in discontinuous-tone content.
Innovation Solution
The method involves dividing coding modes and parameters into two types, with specific Quantization Parameters (QPs) and coding modes for each type, allowing for adaptive coding by calculating and applying a QP difference across multiple coding blocks to optimize bit rate and distortion, and writing these parameters into the bitstream.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single Quantization Parameter (QP) is used for all coding blocks, then the coding process is simple, but the reconstruction quality cannot be adaptively optimized for different content types
Solution Approach 1:
The patent segments coding modes into two distinct types (first type and second type), each associated with different QP values. This segmentation allows the system to apply different quantization strengths to different content regions, achieving adaptive coding without requiring a completely new complex framework
Solution Approach 2:
The patent introduces dynamic QP adjustment by calculating QP differences and applying them to different coding blocks based on their type. The QP is no longer static but dynamically adapted to the content characteristics, enabling optimization for both continuous-tone and discontinuous-tone content
2Productivity
If high QP is used to reduce bit rate, then compression efficiency improves, but distortion in discontinuous-tone content increases significantly
Solution Approach 1:
The patent applies local quality by using different QP values for different coding block types. First type coding blocks (suitable for continuous-tone content) use one QP value, while second type coding blocks (suitable for discontinuous-tone content) use another QP value, ensuring each region receives appropriate quality treatment
Solution Approach 2:
The patent changes the QP parameter dynamically based on coding block type. By calculating QP differences and applying them selectively, the system optimizes the balance between compression efficiency and reconstruction quality for different content characteristics
3Manufacturing precision
If low QP is used to maintain high reconstruction quality, then discontinuous-tone content distortion is minimized, but bit rate increases due to continuous-tone content requirements
Solution Approach 1:
The patent applies local quality by using different QP values for different coding block types. First type coding blocks (suitable for continuous-tone content) use one QP value, while second type coding blocks (suitable for discontinuous-tone content) use another QP value, ensuring each region receives appropriate quality treatment
Solution Approach 2:
The patent changes the QP parameter dynamically based on coding block type. By calculating QP differences and applying them selectively, the system optimizes the balance between compression efficiency and reconstruction quality for different content characteristics
4Adaptability or versatility
If different QP values are applied to different coding blocks, then adaptive optimization is achieved, but the coding and decoding complexity increases
Solution Approach 1:
The patent segments coding modes into two distinct types (first type and second type), each associated with different QP values. This segmentation allows the system to apply different quantization strengths to different content regions, achieving adaptive coding without requiring a completely new complex framework
Data Source
AI summary
Provided are a method and device for coding an image, a method and device for decoding an image. The method for coding the image includes that: coding mode parameters and parameter groups of one coding block are divided into multiple types of coding mode parameters and parameter groups corresponding to the multiple types of coding mode parameters according to a specified rule respectively; Quantization Parameters (QPs) included in the multiple types of coding mode parameters are determined according to a preset target bit rate; a QP of the coding block is determined according to reconstruction quality for the coding block; a coding mode parameter to be used is selected from the multiple types of coding mode parameters according to the QP of the coding block, a parameter group corresponding to the selected coding mode parameter is set, and a QP difference is calculated; and the coding mode parameter, the parameter group used by the coding block and the QP difference are written into a video bitstream.