SSIM-Based Bit Allocation for Perceptual Video Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current bit allocation methods in video encoding primarily focus on Mean Square Error (MSE) optimization, which is not perceptually optimal, whereas Structural Similarity (SSIM) index is more correlated with human visual perception, but lacks corresponding rate and distortion models.
Innovation Solution
A method and apparatus for SSIM-based bit allocation, where a processor estimates the SSIM-based distortion model parameters to determine optimal bit allocation, minimizing overall SSIM distortion and improving human visual quality by allocating bits based on SSIM estimations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If Mean Square Error based bit allocation is used, then the encoding process is simple and computationally efficient, but the perceptual video quality is not optimized
Solution Approach 1:
The patent changes the distortion metric parameter from MSE to SSIM, introducing a new rate-distortion model that uses SSIM index to guide bit allocation. This parameter change enables perceptual optimization while maintaining computational feasibility through the proposed SSIM-based rate-distortion model.
Solution Approach 2:
The patent introduces an intermediary SSIM-based rate-distortion model that bridges the gap between computational efficiency and perceptual quality. This model acts as a mediator, providing a practical framework for bit allocation that incorporates SSIM measurements without requiring complex computational processes.
2Manufacturing precision
If SSIM index is used for quality assessment, then perceptual quality is improved, but no rate and distortion models are available for bit allocation
Solution Approach 1:
The patent establishes new rate and distortion model parameters specifically for SSIM-based bit allocation. By defining these parameters and their relationships, the patent enables practical bit allocation using SSIM while avoiding the complexity of fully perceptual optimization models.
3Power
If bits are allocated to minimize MSE distortion, then computational cost is low, but human visual perception is not optimized
Solution Approach 1:
The patent changes the optimization criterion from MSE minimization to SSIM-based distortion minimization. This parameter change redirects bit allocation toward regions that matter most for human perception, achieving better visual quality without excessive computational cost.
Solution Approach 2:
The patent applies local quality optimization by allocating bits based on SSIM distortion characteristics of different video regions. Regions with higher SSIM distortion (more important for perception) receive more bits, while regions with lower distortion receive fewer bits, optimizing the balance between computational cost and visual quality.
Data Source
AI summary
An embodiment includes a method and an encoder for SSIM-based bits allocation. The encoder includes a memory and a processor utilized for allocating bits based on SSIM, wherein the processor estimates the model parameter of SSIM-based distortion model for the current picture and determines allocates bits based on the SSIM estimation.