Neural Network Video Artifact Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression and decompression methods, such as MPEG, introduce blocky artifacts due to data loss and mismatch between adjacent blocks, which are exacerbated by unknown scaling or shifting during processing, and are further complicated by noise from detailed scenes, real image features, and filter-generated noise, making it difficult to accurately detect and correct these artifacts.
Innovation Solution
A method using a neural network system to analyze video signal characteristics over multiple frames to detect regularly occurring blocky structures, determine their position and scaling, and apply a smoothing filter to correct the artifacts, while distinguishing between true blocky structures and noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a smoothing filter is applied across block transitions to eliminate blockiness, then the visibility of blocky artifacts is reduced, but the system becomes more complex and may be confused by noise from detailed scenes, real image features, and filter-generated noise
Solution Approach 1:
A neural network is introduced as an intermediary system between the video signal and the smoothing filter. The neural network analyzes video frames to detect blocky artifacts and their characteristics (position, size, orientation) before the smoothing filter is applied, enabling the filter to operate more intelligently and reduce confusion from noise and false patterns
Solution Approach 2:
The system continuously monitors video frames and uses the neural network's detection results to adjust the smoothing filter's operation. Feedback from the neural network's analysis of artifact patterns allows the system to adaptively apply smoothing only where needed, distinguishing true blocky artifacts from noise and real image features
2Loss of information
If compression and decompression are performed using block-by-block schemes, then data loss is minimized, but visible blocky artifacts are introduced due to mismatch between adjacent blocks
Solution Approach 1:
The neural network performs preliminary analysis of the video signal to detect blocky artifacts before the smoothing filter is applied. This preliminary detection allows the system to prepare correction data and adjust the smoothing parameters in advance, ensuring that the blocky artifacts are corrected effectively while maintaining the benefits of block-by-block compression
3Adaptability or versatility
If unknown scaling or shifting is applied to the block array during compression, then adaptability to different compression schemes is improved, but the position and characteristics of blocky artifacts become unpredictable and difficult to detect
Solution Approach 1:
The neural network is designed to dynamically adapt to different block positions, sizes, and orientations caused by unknown scaling or shifting operations. The system continuously learns and adjusts its detection parameters based on the actual artifact patterns observed in the video signal, enabling precise detection regardless of the compression scheme applied
Solution Approach 2:
The system changes its detection parameters dynamically based on the observed video data. The neural network adjusts its sensitivity thresholds, detection criteria, and smoothing filter parameters according to the actual blocky artifact characteristics detected in each frame, allowing it to handle various compression schemes effectively
4Quantity of substance
If noise from detailed scenes, real image features, and filter-generated noise is present, then the richness of the video content is maintained, but the ability to detect blocky structures is compromised
Solution Approach 1:
The neural network serves as an intermediary that filters and processes the video signal to distinguish between true blocky artifacts and noise. It analyzes the statistical characteristics and spatial patterns of potential artifacts, using its trained models to identify genuine compression artifacts while ignoring noise from detailed scenes, real image features, and filter-generated noise
Solution Approach 2:
The system uses the presence of noise and complex patterns as training data to improve the neural network's detection capabilities. The noise from detailed scenes and real image features is leveraged to teach the neural network to distinguish between natural image variations and artificial blocky artifacts, converting what was initially a hindrance into a beneficial training resource
Data Source
AI summary
A method and apparatus are provided for removing regularly occurring visible artifacts in decompressed video images. Firstly a decompressed video signal is received. This is filtered frame-by-frame to extract data related to the artifacts. The thus extracted data is then processed in a neural network processor which has been trained to identify the artifacts in order to produce data identifying their locations. The video signal is then corrected to reduce the effect of the thus identified artifacts.