Video Dither Signal Correlation for Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Decoding of video streams compressed at low bit rates often results in visible artifacts like blockiness and structured noise patterns, which become more noticeable due to the human visual system's sensitivity, and existing noise addition techniques risk creating temporal abnormalities.
Innovation Solution
A method is introduced where noise is added to each pixel in a video stream based on the additive noise of a prior picture, with the noise magnitude correlated to the previous picture's noise, reducing large frame-to-frame differences and improving subjective video quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If random noise is added to the decoded video stream to reduce compression artifacts, then spatial artifacts like blockiness and structured noise patterns are reduced, but large frame-to-frame differences are created that produce noticeable temporal artifacts
Solution Approach 1:
The patent applies preliminary action by generating the dither signal in advance through a linear feedback shift register (LFSR) using a seed value derived from the quantization parameter. This pre-generated dither signal is then added to the decoded video stream, ensuring that noise is introduced in a controlled manner that reduces spatial artifacts while maintaining temporal consistency across frames.
Solution Approach 2:
The patent changes the parameter of noise generation by using a deterministic LFSR-based dither signal instead of purely random noise. The seed value for the LFSR is changed based on the quantization parameter, allowing the noise characteristics to be adjusted while maintaining frame-to-frame consistency. This parameter change ensures that the same pixel positions receive consistent dither values across frames, reducing temporal artifacts.
2Object-affected harmful factors
If a dither signal is added to hide compression artifacts, then human sensitivity to block effects and contouring is reduced, but the complexity of the decoding process increases
Solution Approach 1:
The patent replaces the mechanical/random noise generation system with a deterministic linear feedback shift register (LFSR) system. Instead of using complex random number generators or external noise sources, the LFSR provides a simple, efficient, and deterministic method for generating dither signals. This substitution reduces decoding complexity while maintaining the artifact-hiding benefits of dithering.
Solution Approach 2:
The patent applies self-service by generating the dither signal internally within the decoding process using the quantization parameter already available during decoding. The LFSR uses the seed value derived from the quantization parameter to generate the dither signal, eliminating the need for separate noise generation components or external inputs. This self-contained approach simplifies the overall decoding system.
Data Source
AI summary
A decoding arrangement for decoding pictures in an incoming video stream includes a noise generator for adding a dither signal containing random noise to the pictures after video decoding, to improve the subjective video quality. The noise generator adds noise to each pixel in an amount correlated to additive noise of pixels in a prior picture, either a previously displayed picture (i.e., a previously decoded picture to which noise has been added), or a previously decoded picture.