Multimodal Pixel Prediction for Color-Texture Block Mismatch
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques, such as intra block copy mode, are less effective when there are shape matches but color or texture differences between blocks, leading to inefficiencies in bitstream compression.
Innovation Solution
Implementing multimodal prediction by classifying pixels into groups and deriving a model based on adjacent area differences to generate a prediction block, using a common process for encoding and decoding to improve prediction accuracy and reduce side information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If intra block copy mode is used for video compression, then shape matching between blocks can be exploited, but compression efficiency deteriorates when color or texture differences exist between blocks
Solution Approach 1:
The patent segments the reference block and current block into multiple regions based on color and texture characteristics. By dividing blocks into homogeneous regions, the method can apply different prediction strategies to different segments, improving overall compression efficiency while handling color and texture variations effectively.
Solution Approach 2:
The patent applies local quality by using adaptive prediction modes for different regions within blocks. Instead of uniform prediction across entire blocks, the method adjusts prediction parameters locally based on color and texture properties, allowing accurate representation of regions with different characteristics while maintaining compression efficiency.
2Device complexity
If traditional prediction methods are used, then encoding and decoding processes are simple, but prediction accuracy deteriorates when blocks have color or texture differences
Solution Approach 1:
The patent performs preliminary classification of block regions based on color and texture characteristics before applying prediction methods. This preliminary action allows the system to select appropriate prediction modes in advance, improving prediction accuracy without significantly increasing encoding complexity during the actual prediction process.
Solution Approach 2:
The patent introduces dynamic adaptation by selecting different prediction modes and parameters based on local block characteristics. The system dynamically adjusts prediction strategies for different regions, transitioning between simple and complex methods based on actual content needs, thereby improving accuracy while controlling complexity.
3Measurement precision
If high-resolution video is transmitted, then image quality is improved, but bandwidth requirements increase
Solution Approach 1:
The patent uses block copying techniques where reference blocks are copied and adapted to predict current blocks. By exploiting temporal and spatial redundancies through copying, the method reduces the amount of data that needs to be transmitted while maintaining high-resolution image quality through accurate prediction.
Solution Approach 2:
The patent changes prediction parameters adaptively based on block characteristics, such as adjusting prediction mode, block size, and transformation parameters. This parameter adaptation allows more efficient encoding of different video content types, reducing overall bandwidth requirements while preserving image quality through optimized compression.
Data Source
AI summary
The present disclosure provides a method and non-transitory computer readable medium for predicting pixels of a current block. The method includes classifying pixels of a reference block into a first plurality of groups and classifying pixels of a current block adjacent area into a second plurality of groups. A first model for transforming a first group of the first plurality of groups is derived based on pixels in a first group of the first plurality of groups and pixels in a first group of the second plurality of groups. A prediction block for the current block is generated by applying the first model to pixels in the first group of the first plurality of groups. A compressed bitstream encoded by an encoder or decodable by a decoder using the prediction method is also provided.