Occlusion Region Detection in Image Interpolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing techniques face challenges in accurately detecting occlusion regions in interpolation image frames, especially when objects move between successive frames, leading to inaccuracies in interpolating occlusion information due to invalid Motion Vectors (MVs).
Innovation Solution
A method and apparatus that adjust the range of estimated occlusion regions by using Motion Vectors (MVs) mapped to sub-blocks, comparing temporary MVs with random MVs, and excluding sub-blocks if the difference exceeds a threshold, to improve the detection of occlusion regions in interpolation frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If Motion Vectors (MVs) are used to estimate occlusion regions in interpolation frames, then the detection process can be simplified and processed efficiently, but the accuracy of occlusion region detection deteriorates due to invalid MVs causing incorrect interpolation
Solution Approach 1:
The patent segments the occlusion region detection process into multiple stages: initial estimation using MVs, identification of candidate regions, and refined detection within those candidates. This segmentation allows efficient initial processing while ensuring accurate final detection by focusing computational resources on critical areas.
Solution Approach 2:
The patent applies different processing strategies to different regions: MV-based estimation is applied to general areas for efficiency, while more rigorous detection methods are applied specifically to identified candidate occlusion regions. This local differentiation maintains overall accuracy while preserving processing efficiency.
2Reliability
If the range of occlusion regions is expanded to cover all possible motion areas, then no occlusion region is missed, but the detection precision deteriorates due to including invalid regions
Solution Approach 1:
The patent performs preliminary MV-based estimation to identify candidate occlusion regions before conducting refined detection. This preliminary action narrows down the search space, ensuring that the final detection focuses only on relevant regions, thus maintaining both completeness and precision.
Solution Approach 2:
The patent dynamically adjusts the detection range based on MV data, expanding to include potential occlusion areas identified by motion analysis while contracting to exclude regions where MVs indicate no occlusion. This dynamic adjustment ensures the detection range adapts to actual motion patterns, maintaining reliability without sacrificing precision.
Data Source
AI summary
A method for processing image data corresponding to temporally successive motions of an object, the method including adjusting a range of a first occlusion region which is estimated according to whether position changes occur in sub-blocks forming temporally successive first and second frames among image frames which display the motions of the object, by using Motion Vectors (MVs) mapped to the sub-blocks and detecting a second occlusion region of a third frame which displays the object that moves between the first frame and the second frame, by using the adjusted range of the first occlusion region.