Transform Coefficient Compensation for Quantization Loss in Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently reconstructing transform coefficients, leading to image quality loss due to zeroing out high-frequency coefficients and quantization errors, particularly in advanced codecs like VVC.
Innovation Solution
Implementing vector quantization (VQ) based transform coefficient compensation using a codebook with pretrained codewords, where a search key is formed to find a compensation term for reconstructing blocks based on available information at the encoder and decoder, compensating for losses in transform coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If quantization is applied to transform coefficients for compression, then bitrate is reduced, but image quality deteriorates due to quantization errors
Solution Approach 1:
The patent applies preliminary action by performing compensation before final reconstruction. A search key is formed from available information (quantized transform coefficients, prediction mode, block size) to search a codebook for compensation terms. These compensation terms are applied to correct quantization errors before the final inverse transform, thereby improving image quality without increasing the transmitted bitrate.
2Quantity of substance
If high-frequency transform coefficients are zeroed out for compression, then data volume is reduced, but image quality deteriorates due to loss of high-frequency details
Solution Approach 1:
The patent uses copying by creating a search key that replicates essential information from the quantized transform coefficients, prediction mode, and block size. This search key is then used to search a codebook for compensation terms that copy the missing high-frequency details. The compensation terms effectively reconstruct the lost information without requiring transmission of the original high-frequency coefficients.
3Manufacturing precision
If vector quantization with codebook search is implemented for compensation, then image quality improves, but computational complexity increases
Solution Approach 1:
The codebook is pre-trained and prepared in advance, containing compensation terms for various scenarios. During decoding, the process involves forming a search key from available information and searching the pre-prepared codebook, which is computationally more efficient than performing full optimization during real-time decoding. This preliminary preparation reduces the computational burden during actual image reconstruction.
Solution Approach 2:
The search key acts as an intermediary between the quantized transform coefficients and the compensation terms in the codebook. Instead of directly searching for compensation terms using all available information, the patent creates a compact search key that captures essential features, thereby simplifying the search process and reducing computational complexity while still enabling effective compensation.
Data Source
AI summary
Some aspects of the disclosure provide a method of video decoding. For example, a coded video bitstream is received. The coded video bitstream includes coded information of a current block in a picture. The coded information includes quantized transform coefficients for a residual block of the current block, the residual block is transformed from a spatial domain to transform coefficients in a spectrum domain, the transform coefficients are quantized into the quantized transform coefficients. A search key is formed based on the coded information of the current block. A codebook is searched by using the search key to obtain a compensation term. The codebook includes a plurality of codewords, each codeword of the plurality of codewords includes a key and a predetermined compensation term associated with the key. The current block is reconstructed based on the compensation term.