Image Encoding with Adaptive Coefficient Range for High-Bit Depth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The limitation of coefficient values in VVC encoding to -32768 to 32767 for quantization processing results in reduced arithmetic precision for high-bit depth images, leading to a decrease in image quality without significantly increasing hardware implementation cost.
Innovation Solution
An image encoding and decoding device that adaptively determines the range of possible coefficient values based on bit depth, allowing for high-precision or fixed ranges, enabling flexible arithmetic precision and implementation cost depending on application requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the coefficient value range is limited to -32768 to 32767 for easier implementation, then hardware implementation cost does not increase much, but arithmetic precision is reduced for high-bit depth images
Solution Approach 1:
The patent applies dynamics by making the coefficient value range adjustable rather than fixed. The encoding device can adaptively determine the coefficient value range based on the bit depth of the input image, allowing the system to dynamically switch between a limited range (-32768 to 32767) for easier implementation and an extended range for higher precision when needed.
Solution Approach 2:
The patent changes the parameter of coefficient value range from a fixed constant to a variable that depends on bit depth. By introducing bit depth as a determining factor, the system can adjust the coefficient range parameter to match the precision requirements of different image types, thereby resolving the contradiction between implementation ease and arithmetic precision.
2Measurement precision
If the coefficient value range is extended for high-bit depth images, then arithmetic precision is improved, but hardware implementation cost increases
Solution Approach 1:
The system dynamically adjusts the coefficient value range based on the actual bit depth of the input image. This means that extended precision is only implemented when necessary (for high-bit depth images), avoiding unnecessary hardware complexity for standard bit depth images, thus balancing precision requirements with implementation cost.
Solution Approach 2:
The patent introduces a parameter change mechanism where the coefficient range is adjusted according to bit depth. This allows the system to optimize hardware resource allocation by only enabling extended precision processing when the input image requires it, rather than providing full precision capability for all cases.
3Device complexity
If a fixed coefficient range is used for all bit depths, then implementation is simplified, but image quality deteriorates for high-bit depth images
Solution Approach 1:
The patent makes the coefficient range dynamic rather than fixed, allowing it to adapt to different bit depth requirements. This dynamic adjustment ensures that high-bit depth images receive the extended precision they need for high quality, while low-bit depth images use the simpler fixed range, thus resolving the contradiction between implementation simplicity and image quality.
Solution Approach 2:
The system changes the coefficient range parameter based on bit depth detection. This parameter adaptation allows the encoding device to maintain simplicity for standard images while automatically providing enhanced precision for high-bit depth images, thereby maintaining image quality without universally increasing complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image encoding device includes a prediction unit configured to generate prediction errors being a difference between a predicted image obtained by prediction processing for an input image and the input image, a first transform unit configured to generate first transform coefficients by performing orthogonal transform on the prediction errors, a second transform unit configured to generate second transform coefficients by performing LFNST processing on the first transform coefficients, a quantization unit configured to generate quantization coefficients by performing quantization processing on the second transform coefficients, and an encoding unit configured to encode the quantization coefficients, wherein the encoding unit encodes information indicating whether a range of possible values at least taken by the second transform coefficients is to be a range determined based on a bit depth or a fixed range.