Self-adaptive prediction for multi-layer codec bit rate overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-layer video codecs face inefficiencies in transmitting inverse mapping parameters, leading to high bit rate overhead, especially in low resolution video streaming, as they are not adapted to individual image frames or partitions, resulting in suboptimal rate distortion optimization and coding efficiency, particularly in scenes with changing content or special editing effects.
Innovation Solution
Implementing self-adaptive prediction techniques that generate inverse mapping parameters on a per-image or per-partition basis using temporal and spatial domain analysis, reducing the need for parameter transmission by leveraging statistics from previous frames or partitions, and using mechanisms like recursive least squares and 3D lookup tables for improved prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If inverse mapping parameters are transmitted for each partition in multi-layer video coding, then prediction accuracy is improved, but bit rate overhead increases significantly
Solution Approach 1:
The patent segments the video content into base layer and enhancement layer, with the base layer carrying essential visual information and the enhancement layer carrying residual data. This segmentation allows the system to transmit inverse mapping parameters only for the enhancement layer, reducing overall bit rate overhead while maintaining prediction accuracy where it matters most.
Solution Approach 2:
The patent applies different coding strategies to different parts of the video signal. The base layer uses conventional coding without requiring inverse mapping parameters, while the enhancement layer utilizes adaptive inverse mapping parameters. This local differentiation optimizes the balance between prediction accuracy and bit rate overhead by applying sophisticated parameter transmission only where needed.
2Quantity of substance
If fixed inverse mapping parameters are used for all frames, then transmission overhead is reduced, but coding efficiency deteriorates in scenes with changing content
Solution Approach 1:
The patent introduces dynamic adaptation of inverse mapping parameters for the enhancement layer based on scene content characteristics. The system adapts parameters frame-by-frame or partition-by-partition, allowing coding efficiency to respond to changing content while maintaining reasonable transmission overhead through the layered structure.
Solution Approach 2:
The base layer serves the enhancement layer by providing a reconstructed signal that the enhancement layer uses as a reference for generating adaptive inverse mapping parameters. This self-service mechanism allows the system to generate optimization parameters dynamically without requiring external input, balancing overhead and efficiency automatically.
3Measurement precision
If partition-based prediction is used to improve visual quality, then prediction precision is enhanced, but parameter overhead becomes proportional to the number of partitions
Solution Approach 1:
The patent adds a layer dimension to the conventional single-layer coding structure. By introducing the enhancement layer with adaptive inverse mapping parameters, the system achieves partition-based prediction precision without proportionally increasing overhead, as the parameters are optimized across the layer dimension rather than purely within each partition.
Solution Approach 2:
The patent changes the parameters of the enhancement layer adaptively based on content characteristics, rather than using fixed parameters for all partitions. This parameter adaptation allows the system to maintain high prediction precision only where necessary, reducing the effective number of parameters that need to be transmitted while preserving visual quality.
Data Source
AI summary
Relatively low dynamic range images or image partitions are converted into relatively high dynamic range images or image partitions that comprise reconstructed pixel values having a higher dynamic range than pixel values of the relatively low dynamic range images. Information relating to reconstructed pixel values of the relatively high dynamic range images and pixel values of the relatively low dynamic range images is collected. Prediction parameters are derived from the collected information. A predicted image or image partition is predicted from a relatively low dynamic range image or image partition based on the prediction parameters and comprises predicted pixel values having the higher dynamic range than pixel values of the relatively low dynamic range image or image partition.