Viewpoint Disparity Image Rendering via Reprojection and Occlusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Embedded devices with limited processing capability and high power consumption struggle to match the graphics processing capabilities of workstations, making it difficult to render high-quality images with binocular or viewpoint disparity efficiently.
Innovation Solution
A method and device that perform rendering on first and second images with binocular or viewpoint disparity by generating depth information, determining corresponding pixels, and performing reprojection to select and render pixels based on an occlusion map, reducing unnecessary calculations through tile-based rendering and depth testing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If rendering is performed on all pixels of the second image independently, then complete image rendering is achieved, but computational load increases significantly
Solution Approach 1:
The patent performs preliminary rendering on the first image to generate depth information and occlusion maps before rendering the second image. This preliminary action identifies which pixels of the second image can be obtained through reprojection from the first image, avoiding redundant rendering calculations and significantly reducing computational load while maintaining complete image rendering quality
Solution Approach 2:
The patent creates a copy of depth information from the first image and uses it to determine corresponding pixels in the second image through reprojection. By copying and reusing depth data, the system avoids performing independent rendering calculations for all pixels of the second image, thereby improving rendering efficiency without sacrificing image quality
2Measurement precision
If depth information is generated for all pixels, then accurate reprojection is achieved, but processing time and power consumption increase
Solution Approach 1:
The patent applies local quality by generating depth information and processing pixels only where necessary. Through occlusion maps and selective reprojection, the system processes only the specific regions and pixels that require depth information for accurate rendering, rather than uniformly processing all pixels. This localized approach maintains depth information accuracy for critical areas while reducing overall processing time and power consumption
3Productivity
If tile-based rendering with scheduling is implemented, then processing efficiency is improved, but device complexity increases
Solution Approach 1:
The patent divides the rendering process into discrete tiles and implements scheduling for each tile based on occlusion map data. By segmenting the second image into tiles and determining which tiles require rendering versus which can be obtained through reprojection, the system improves processing efficiency through parallel processing and resource optimization. The segmentation approach manages complexity by breaking down the rendering task into smaller, independently schedulable units
Data Source
AI summary
A device and method performing rendering on first and second images having binocular or viewpoint disparity are provided. The method includes determining some pixels of the second image that correspond to pixels of the first image, performing reprojection to determine, based on rendering results of the first image, color values for the pixels of the second image that are determined to correspond to the pixels of the first image, and selectively rendering remaining pixels of the second image based on the reprojection.