| Summary: | 碩士 === 大葉大學 === 資訊工程學系碩士班 === 99 === In this paper, we proposed a hierarchical disparity estimation scheme using
region matching. The proposed method is composed of hierarchical structure,
region-based disparity estimation, and disparity refinement. In the hierarchical
structure, three procedures, color transform, edge detection and sub-sample, are
performed. To perform disparity estimation, a cost function based on normalized cross
correlation as well as color and spatial features is developed and used as a
measurement in the proposed disparity estimation method. And get a high resolution
depth map. In disparity refinement, the disparity values in occlusion regions are
estimated and patch outlier.
According to experimental results, our proposed scheme can perform disparity
estimation well.
|
|---|
