Equirectangular Projection Video Encoding Quantization Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing moving image encoding and decoding technologies fail to appropriately set and predict quantization parameters for equirectangular projection formats, leading to reduced performance at screen edges and excessive quantization, which impairs the sense of presence in virtual reality experiences.
Innovation Solution
A moving image encoding and decoding device that adjusts quantization parameters based on the position within the equirectangular projection format, using a correction process to account for the varying pixel density and aspect ratio, thereby optimizing quantization and reducing block noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional quantization parameter prediction is used for equirectangular projection format, then encoding complexity is reduced, but image quality deteriorates at screen edges due to excessive quantization
Solution Approach 1:
The patent applies different quantization parameter prediction strategies to different regions of the equirectangular projection frame. Specifically, it identifies that the upper and lower edge regions (corresponding to polar areas in the spherical projection) require different handling compared to the central region. By applying local quality adjustment through region-specific quantization parameter prediction, the patent resolves the contradiction by maintaining appropriate image quality in edge regions while keeping the overall encoding process manageable.
Solution Approach 2:
The patent changes the quantization parameter prediction approach based on the spatial position within the equirectangular projection frame. It introduces position-dependent quantization parameter adjustment, where the prediction method or parameters are modified according to whether the current block is located in the upper edge region, lower edge region, or central region. This parameter change enables the system to adapt to the varying visual importance and distortion characteristics across different frame regions.
2Device complexity
If uniform quantization parameter is applied across the entire frame, then device complexity is reduced, but block noise increases in regions with varying pixel density
Solution Approach 1:
The patent implements local quality control by applying different quantization parameters to different spatial regions of the frame. It specifically addresses the block noise problem in upper and lower edge regions by using region-aware quantization parameter prediction, which adapts to the varying pixel density characteristics. This local differentiation reduces block noise in critical regions while maintaining manageable device complexity through systematic region classification.
Solution Approach 2:
The patent segments the frame into distinct regions (upper edge region, lower edge region, and central region) for differential quantization parameter application. By dividing the frame into these segments and applying appropriate quantization strategies to each, the system effectively manages block noise in high-density regions while maintaining overall complexity at acceptable levels through the structured segmentation approach.
3Productivity
If quantization parameter prediction uses adjacent blocks, then encoding efficiency is improved, but performance deteriorates at screen edges where adjacent blocks may not represent the same visual importance
Solution Approach 1:
The patent modifies the conventional adjacent-block prediction approach by introducing region-aware selection. When predicting quantization parameters for blocks in the upper or lower edge regions, the system selectively uses adjacent blocks that are also in the same region or appropriately positioned, rather than blindly using any adjacent block. This local quality consideration ensures that the prediction reflects the actual visual importance and distortion characteristics of the current region, improving quantization parameter accuracy at screen edges while maintaining encoding efficiency.
Data Source
Figure 1~3
Figure 4~5
Figure 6~7
AI summary
A moving image encoding device that performs encoding by converting and quantizing a prediction residual difference obtained by applying an interframe prediction or an intraframe prediction to each encoding tree block, which is a divided unit of a frame given by means of equirectangular projection, for each encoding block. The moving image encoding device includes: a calculation unit that calculates a quantization parameter to be applied to the encoding block on the basis of a pixel value of a pre-coding block and a position in the frame; a prediction unit that obtains a prediction value by predicting the quantization parameter on the basis of a quantization parameter of a left neighboring, and upper neighboring block of the encoding block, or an already encoded block; a correction unit that corrects the prediction value and obtains a correction value when a condition is satisfied where the encoding block is positioned at a left end of the frame and the left neighboring block and the upper neighboring block belong to encoding tree blocks that are different from the encoding tree block to which the encoding block belongs, and that obtains the prediction value as a correction value when the condition is not satisfied; and an encoding unit that encodes a difference between the correction value and the quantization parameter.