단일 카메마로 획득한 영상 쌍의 평면 호모그래피를 이용한 평면 검출

Abstract

Detecting a planar surface in real world(3D) is important to recognize environment for mobile robot. Especially, planar surface in indoor environment structured plane can be used for information to navigate safely and recognize the plane for task. This paper proposes a method to detect plane regions using the affine homography from image pair. First, using image pair from monocular sequence, corresponding point is obtained by SURF. At least three corresponding points are used to calculate the affine homography that represents a plane in the image. That is used for detecting points belonging to the same plane based on the reprojection error. Using points lying on the same plane first, we find new affine homography to minimize the reprojection error. When the reprojection error can not minimize any longer, the homography will be optimum planar homography. Experiment results are performed in different sequence of indoor image pair from wheeled mobile robot and results are shown in section 4.