Noise Adaptive 3D Composite Video Filtering for Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video filtering techniques are inefficient in detecting and reducing cross-chroma and cross-luma artifacts, particularly in low-end applications, due to high computational and memory costs, and fail to accurately estimate noise levels in high-texture or high-motion scenarios, and cannot correctly filter 'circular sweep zone plate' patterns or differentiate between strong and weak dot crawl.
Innovation Solution
A noise adaptive 3-dimensional composite noise reduction technique that filters digital video by checking for artifacts using an adjustable threshold, applying motion-compensated filtering to blocks with noise, and using a cross spatio-temporal sum-of-absolute-differences approach to detect cross-luma artifacts, while filtering differently for strong and weak dot crawl.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion compensation is used for noise level measurements, then measurement reliability is improved, but computational cost and memory bandwidth increase
Solution Approach 1:
The video signal is divided into blocks for independent processing. Noise level measurements are performed on individual blocks rather than entire frames, reducing computational complexity while maintaining accuracy where needed.
Solution Approach 2:
Motion compensation is applied selectively only to blocks where it provides meaningful improvement in measurement accuracy, rather than uniformly across the entire video signal. This local application reduces overall computational cost while preserving measurement reliability in critical regions.
2Device complexity
If spatial based methods are used for noise detection, then implementation simplicity is improved, but ability to distinguish high texture from high noise deteriorates
Solution Approach 1:
The detection method uses feedback from multiple sources including temporal information and motion data to refine noise estimates. This feedback mechanism allows the simple spatial method to achieve better discrimination between texture and noise by iteratively improving measurements.
Solution Approach 2:
The patent combines multiple detection approaches (spatial, temporal, motion-based) into a composite detection system. This composite method leverages the strengths of each individual approach while mitigating their weaknesses, achieving accurate texture-noise discrimination without excessive complexity.
3Measurement precision
If existing shimmer detection methods are used, then cross-chroma and cross-luma detection capability is improved, but computational efficiency and bandwidth efficiency deteriorate
Solution Approach 1:
Artifact detection is performed on segmented blocks rather than entire frames, enabling parallel processing and reducing the computational burden per processing unit. This segmentation maintains detection accuracy while improving overall processing efficiency.
Solution Approach 2:
The patent applies filtering and detection operations selectively only to blocks where artifacts are detected, rather than processing the entire video signal uniformly. This partial action approach maintains artifact detection capability while significantly reducing computational and bandwidth requirements.
4Device complexity
If uniform filtering is applied to all blocks, then implementation simplicity is improved, but ability to handle different artifact strengths deteriorates
Solution Approach 1:
The filtering strength is adapted locally to each block based on detected artifact characteristics. Blocks with strong dot crawl receive different filtering treatment than blocks with weak dot crawl, allowing the system to handle varying artifact strengths effectively while maintaining reasonable implementation complexity.
Solution Approach 2:
The filtering parameters are made dynamic and adaptive rather than static and uniform. The filter automatically adjusts its strength based on real-time detection of artifact characteristics, enabling effective handling of both strong and weak artifacts without requiring complex manual configuration.
Data Source
AI summary
A method for filtering digital video is disclosed. The method generally includes the steps of (A) checking a plurality of original blocks in an original field for a plurality of artifacts, wherein detection of the artifacts is based on an adjustable threshold, (B) first filtering each of the original blocks having at least one of the artifacts to remove the at least one artifact and (C) second filtering each of the original blocks lacking at least one of the artifacts to remove noise, wherein the second filtering is (i) motion compensated and (ii) adaptive to a respective noise level in each of the original blocks.