CNN-Based Video Filtering for Quantization and Blocking Artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing data volume and complexity of video content, particularly in high-capacity games and 360-degree videos, necessitates a more efficient compression technique to reduce hardware resource consumption and mitigate quantization errors and blocking artifacts in video encoding and decoding processes.
Innovation Solution
Applying a convolutional neural network (CNN)-based filter to video encoding and decoding operations, utilizing a quantization parameter map and block partition map to enhance reconstructed pictures, and performing CNN-based intra- and inter-prediction to improve prediction accuracy while maintaining decoding complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If video data is compressed using traditional encoders, then data volume is reduced, but quantization errors and blocking artifacts increase
Solution Approach 1:
The patent replaces traditional mechanical filtering systems with a neural network-based system. The neural network learns optimal filtering operations during training and applies them during decoding, substituting conventional signal processing methods with machine learning-based approaches that adaptively reduce quantization errors and blocking artifacts while maintaining compression efficiency
Solution Approach 2:
The patent changes the parameters of the filtering process by using a neural network that learns optimal filter coefficients and processing parameters during training. The network dynamically adjusts filtering strength and characteristics based on the input picture characteristics, enabling adaptive quality enhancement that responds to local picture content rather than applying fixed filtering parameters
2Manufacturing precision
If CNN-based filter is applied to mitigate quantization errors, then picture quality is improved, but decoding complexity increases
Solution Approach 1:
The patent performs the complex neural network training and filter optimization in advance during an encoding phase. The trained neural network model and its parameters are stored and reused during decoding, so the computationally intensive work is done beforehand rather than during real-time playback, reducing instantaneous decoding complexity while maintaining quality improvement
Solution Approach 2:
The patent uses a trained neural network model that captures the essence of complex filtering operations. Instead of implementing multiple traditional filtering passes, the system uses the trained network as a compact representation that can be applied efficiently during decoding, copying the learned knowledge into a reusable computational structure that simplifies the decoding process
Data Source
AI summary
The present disclosure relates to video encoding or decoding and, more specifically, to an apparatus and a method for applying an artificial neural network (ANN) to video encoding or decoding. The apparatus and the method of the present disclosure are characterized by applying a CNN-based filter to a first picture and at least one of a quantization parameter map and a block partition map to output a second picture.