中国科学院重庆绿色智能技术研究院机构知识库

QQ客服 官方微博 反馈留言

	Invariant matching method for different viewpoint angle images
	Chen, Min1; Shao, Zhenfeng1; Li, Dongyang1; Liu, Jun2
	2013
摘要	In recent years, many methods have been put forward to improve the image matching for different viewpoint images. However, these methods are still not able to achieve stable results, especially when large variation in view occurs. In this paper, an image matching method based on affine transformation of local image areas is proposed. First, local stable regions are extracted from the reference image and the test image, and transformed to circular areas according to the second-order moment. Then, scale invariant features are detected and matched in the transformed regions. Finally, we use epipolar constraint based on the fundamental matrix to eliminate wrong corresponding pairs. The goal of our method is not to increase the invariance of the detector but to improve the final performance of the matching results. The experimental results demonstrate that compared with the traditional detectors the proposed method provides significant improvement in robustness for different viewpoint images matching in the 2D scene and 3D scene. Moreover, the efficiency is greatly improved compared with affine scale invariant feature transform (Affine-SIFT). (c) 2012 Optical Society of America
DOI	10.1364/AO.52.000096
发表期刊	APPLIED OPTICS
ISSN	1559-128X
卷号	52期号:1页码:96-104
通讯作者	Shao, ZF (reprint author), Wuhan Univ, State Key Lab Informat Engn Surveying Mapping & R, 129 Luoyu Rd, Wuhan 430079, Hubei, Peoples R China.
收录类别	SCI
WOS记录号	WOS:000313745800060
语种	英语