Segment-Based Matching for Planar Region Extraction
Jae-Ho Yun, Rae‐Hong Park · ITC-CSCC :International Technical Conference on Circuits Systems, Computers and Communications · 2007
This paper presents a planar region extraction method using a color segment-based matching approach. Planar regions in the three-dimensional space are geometric information for reconstruction of the scene structure and have been used to reconstruct the scene or synthesize the view. This paper focuses on accurate extraction of planar regions from a pair of images and proposes a segment-based matching method. The proposed method starts with an iterative random sample consensus based homography estimation using point correspondences. Then, color segmentation is performed to split each image into a number of homogeneous color segments. If some segments correspond to the same homography, it means that those segments lie on the same planar region. Thus, this paper employs a segment-based matching to determine the homography corresponding to each color segment for planar region extraction. By the proposed algorithm, each color segment is correctly classified into one of corresponding planar regions. Experimental results with a number of test image pairs show the effectiveness of the proposed method.