Transform Coefficient Sign Prediction for Lower-Bitrate Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently compressing video data while maintaining video quality, particularly in the prediction of transform coefficient signs during block-based video coding.
Innovation Solution
Implementing a template-based hypothesis generation scheme for transform coefficient sign prediction on the video decoder side, which involves selecting candidate transform coefficients, applying a hypothesis generation scheme, determining predicted signs, and updating dequantized coefficients based on received sign signaling bits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional video coding methods are used for transform coefficient sign prediction, then the decoding process is simple, but the video compression efficiency and quality are insufficient
Solution Approach 1:
The patent applies preliminary action by performing sign prediction before final coefficient reconstruction. The decoder generates predicted signs for transform coefficients using template-based hypothesis generation from neighboring block signs, then applies these predictions during the decoding process to improve reconstruction accuracy before outputting the final video data.
Solution Approach 2:
The patent introduces an intermediary mechanism by using sign prediction as a intermediate step between receiving quantized coefficients and reconstructing the final video block. The predicted signs act as a mediator that refines the coefficient values before they are used in the inverse transform, improving overall video quality without requiring complete re-encoding.
2Measurement precision
If more bits are allocated for sign information, then the accuracy of transform coefficient reconstruction improves, but the bit rate increases
Solution Approach 1:
The patent implements feedback by using signs from previously decoded neighboring blocks to predict the signs of current block coefficients. This feedback mechanism allows the decoder to infer sign information from already-reconstructed data, reducing the need to transmit explicit sign bits while maintaining reconstruction accuracy through the predictive relationship between adjacent blocks.
Solution Approach 2:
The patent applies copying by replicating sign patterns from neighboring blocks to the current block. The template-based hypothesis generation copies sign information from spatially adjacent blocks and uses it to predict the current block's coefficient signs, thereby avoiding the need to transmit redundant sign data for each coefficient individually.
3Measurement precision
If a template-based hypothesis generation scheme is applied, then the sign prediction accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the sign prediction process into distinct stages: hypothesis generation from multiple templates, hypothesis evaluation, and final sign determination. This segmentation allows the complex prediction task to be broken into manageable steps, each operating on specific candidate hypotheses rather than evaluating all possible sign combinations simultaneously.
Solution Approach 2:
The patent implements partial action by generating a limited set of candidate hypotheses from specific template patterns rather than evaluating all possible sign configurations. The method selectively applies a predetermined number of hypotheses based on template matching, performing sufficient prediction action to achieve accuracy while avoiding the excessive computational burden of exhaustive search.
Data Source
AI summary
Implementations of the disclosure provide a video decoding apparatus and method for transform coefficient sign prediction on a video decoder side. The method may include generating a plurality of candidate hypotheses for a set of candidate transform coefficients associated with a transform block of a video frame from a video. The method may further include selecting a hypothesis from the plurality of candidate hypotheses as a set of predicted signs for the set of candidate transform coefficients, wherein the hypothesis is selected based on a cost function calculated by extrapolating neighboring samples of the transform block in an extrapolation direction determined based on a dominant gradient direction. The method may also include estimating original signs for the set of candidate transform coefficients based on the set of predicted signs and a sequence of sign signaling bits received from a video encoder.