CNN-Based Video Codec Filters 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.
Innovation Solution
Applying a convolutional neural network (CNN)-based filter to video encoding and decoding processes, utilizing quantization parameter maps and block partition maps 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 to reduce data volume, then hardware resource consumption is reduced, but quantization errors and blocking artifacts increase
Solution Approach 1:
The patent replaces traditional mechanical filtering systems with an artificial neural network-based filter. The neural network learns optimal filtering operations during training and applies learned patterns during decoding, substituting conventional signal processing mechanisms with intelligent, adaptive processing that reduces artifacts while maintaining compression efficiency
Solution Approach 2:
The patent changes the operational parameters of the filtering process by using learned filter coefficients and adaptive filtering strength control. The neural network dynamically adjusts filtering parameters based on local picture characteristics, allowing optimized balance between artifact removal and detail preservation without fixed parameter constraints
2Manufacturing precision
If CNN-based filter is applied to enhance reconstructed pictures, then picture quality improves, but decoding complexity increases
Solution Approach 1:
The patent performs the complex filtering operations in advance during the training phase, where the neural network learns optimal filtering strategies. During actual decoding, the pre-trained network applies learned patterns efficiently without requiring complex real-time computations, thus reducing operational complexity while maintaining quality improvements
Solution Approach 2:
The patent uses the neural network to learn and copy effective filtering patterns from training data. The network captures successful artifact removal strategies during training and reproduces them during decoding, allowing complex filtering behavior to be replicated through learned representations rather than explicit complex algorithms
3Measurement precision
If video resolution and frame rate are increased to meet growing demand, then video content quality improves, but data volume to be compressed increases rapidly
Solution Approach 1:
The patent changes the compression efficiency parameter by introducing neural network-based filtering that removes artifacts more effectively. This allows achieving the same perceptual quality at lower bitrates or maintaining quality at higher resolutions, effectively changing the quality-to-data-volume ratio through intelligent processing
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.