Occlusion Region Extraction for Multi-View Image Data Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Three-dimensional (3D) video systems face challenges in efficiently storing and transmitting data due to the large amount of data required for multi-view images, particularly in acquiring and encoding color images across different viewpoints, with a need to address occlusion region information caused by view differences.
Innovation Solution
An image processing apparatus and method that extracts occlusion regions from color and depth images using differences between warped and target views, applies backward warping, and generates occlusion region images with side offsets to reduce data storage and transmission by encoding occlusion region information with the reference view images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If color images of multiple views are substantially acquired and stored, then the quality and completeness of 3D image data is improved, but the amount of data to be stored and transmitted increases significantly
Solution Approach 1:
The patent extracts and processes only the occlusion region information from the reference view image, separating it from the complete image data. This allows the system to store and transmit minimal data (occlusion masks and depth maps) while still enabling reconstruction of multi-view images, thus reducing data quantity while maintaining reliability
Solution Approach 2:
The patent uses depth maps and occlusion masks as simplified copies that represent the essential structural information needed for multi-view reconstruction. Instead of storing actual color images of all views, the system uses these compact representations that can be used to generate the full images when needed
2Measurement precision
If occlusion region information is appropriately acquired and processed, then the accuracy of 3D image reconstruction is improved, but the processing complexity increases
Solution Approach 1:
The patent segments the image processing into distinct components: depth map generation, occlusion mask extraction, and backward warping. By dividing the complex task of occlusion region processing into these manageable segments, the system achieves accurate extraction while keeping each processing stage relatively simple and modular
Solution Approach 2:
The patent performs preliminary processing to generate depth maps and occlusion masks from the reference view before actual multi-view reconstruction is needed. This preliminary action prepares the essential data structures in advance, making the subsequent reconstruction process more accurate and efficient
Data Source
AI summary
Provided is an image processing apparatus. In the case of a multi-view image, the image processing apparatus may warp a color image and/or a depth image of a reference view to correspond to a target view. An occlusion region may be extracted by comparing the warped color image and/or depth image of the reference view with a color image and/or a depth image of the target view. The image processing apparatus may encode the extracted occlusion region information without loss, and decrease an amount of data in the multi-view image.