Content Adaptive Video Decoding via Scene-Specific Decoder Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression techniques fail to provide acceptable performance across a wide range of video content characteristics, as they are optimized for specific types of scenes rather than adapting to the diverse characteristics of video content, leading to inefficiencies in encoding and decoding.
Innovation Solution
A content adaptive video codec that segments video content into portions optimized for specific encoding and decoding techniques based on characteristics, using a plurality of encoders and decoders associated with different models, allowing for optimal coding and decoding schemes for each portion of the video content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a generic video codec is used to handle diverse video content, then the device complexity is reduced, but the compression quality and efficiency deteriorate for specific scene types
Solution Approach 1:
The video content is segmented into different scene types (e.g., video telephony scenes, movie scenes, news scenes) and each segment is decoded by a specialized decoder optimized for that scene type. This segmentation allows the system to achieve high compression quality for each specific scene type while maintaining manageable device complexity through modular decoder design.
Solution Approach 2:
Different parts of the video content (different scene types) are processed with different quality levels and decoding approaches. Each scene type receives specialized processing tailored to its characteristics, such as using motion-compensated prediction for video telephony scenes and advanced intra-prediction for movie scenes, thereby optimizing compression quality locally for each scene type.
2Manufacturing precision
If a specialized decoder is used for each scene type, then the compression quality is improved, but the device complexity increases
Solution Approach 1:
The system employs a universal decoder architecture that can function in multiple modes. A single decoder device can be configured to handle different scene types by selecting appropriate decoding algorithms and parameters, rather than requiring separate dedicated decoders for each scene type. This multi-functionality reduces device complexity while maintaining specialized processing capabilities.
Solution Approach 2:
The decoder is designed to dynamically adapt its processing mode based on the input scene type. The system can switch between different decoding algorithms and complexity levels in real-time depending on the detected scene characteristics, allowing a single decoder to perform the work of multiple specialized decoders while maintaining optimal compression quality for each scene type.
3Productivity
If video content is processed scene-by-scene with different techniques, then the compression efficiency is improved, but the processing time and delay increase
Solution Approach 1:
The system performs preliminary classification of video content into scene types before decoding. By identifying the scene type early in the processing pipeline, the appropriate decoding algorithm can be selected and prepared in advance, avoiding time-consuming algorithm switching during decoding and minimizing processing delay while maintaining high compression efficiency.
Solution Approach 2:
The system uses pre-configured decoding profiles and parameter sets for different scene types that can be quickly loaded and applied. Instead of computing optimal decoding parameters in real-time, the system copies and applies pre-optimized parameter sets corresponding to the detected scene type, significantly reducing processing time while maintaining compression efficiency.
Data Source
AI summary
A method and computer-readable medium for content adaptively decoding video content is disclosed. A computer-readable medium stores instructions for controlling a computing device to decode a bitstream encoded via a plurality of encoders. The bitstream is divided into portions and each portion has an associated model chosen from a plurality of predefined models. The instructions comprise routing each portion via an input switch to one of a plurality of decoders based on the associated model of the portion.