Intra Prediction Mode Derivation for Efficient Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face challenges in achieving high coding efficiency due to increasing image sizes, resolutions, and frame rates, necessitating improved methods for deriving intra prediction modes in video compression.
Innovation Solution
A video coding method and apparatus that derive the intra prediction mode of a current block using pre-restored, neighboring reference sample values and generate a prediction block based on calculated gradient histograms, incorporating a prediction mode derivation flag and gradient calculation areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If video data is stored or transmitted without compression processing, then image quality is preserved, but hardware resources including memory are excessively consumed
Solution Approach 1:
The patent applies parameter changes by transforming video data from spatial domain to frequency domain through transform processing (e.g., DCT transform). This changes the representation parameters of the data, allowing energy compaction where most important information is concentrated in fewer coefficients, thereby reducing the quantity of data to be stored or transmitted while preserving image quality through selective quantization of less important coefficients.
2Reliability
If image size, resolution, and frame rate are increased, then image quality is improved, but coding efficiency deteriorates and data amount increases
Solution Approach 1:
The patent employs parameter changes through advanced transform techniques (e.g., integer transform, floating-point transform) and adaptive quantization that dynamically adjust transformation parameters based on content characteristics. This allows efficient compression of high-resolution, high-frame-rate video by adapting the transformation and quantization parameters to the specific content, maintaining image quality while improving coding efficiency.
Solution Approach 2:
The patent applies dynamics through adaptive prediction modes and content-dependent processing where the coding parameters are dynamically adjusted based on the local content characteristics. Different prediction modes (intra prediction, inter prediction) are selected and applied adaptively to different regions, allowing the system to handle varying image sizes, resolutions, and frame rates efficiently while maintaining high image quality.
3Ease of operation
If intra prediction mode is parsed from bitstream, then decoding complexity is reduced, but coding efficiency is insufficient for high resolution video
Solution Approach 1:
The patent applies preliminary action by performing prediction mode derivation and selection during the encoding process based on content analysis, and then encoding only the necessary residual information or mode indices. This preliminary determination of optimal prediction modes allows the decoder to reconstruct high-resolution video with reduced complexity, as the heavy lifting of mode selection is done in advance during encoding based on gradient analysis and content characteristics.
Data Source
AI summary
A method and an apparatus for video coding using derivation of intra prediction modes are disclosed. The video coding method and the video coding apparatus derive the intra prediction mode of a current block using pre-restored, neighboring reference sample values and generate a prediction block of the current block based on the derived prediction mode.