Singular value decomposition based image matching

This paper presents a simple and effective method for matching two uncalibrated images. Corner points are firstly extracted as interest points in the two images. Each interest point is assigned one dominant orientation. The initial set of point matches is then obtained by singular value decomposition of a correspondence strength matrix. A new expression of this matrix is introduced to handle more complicated imaging conditions. Each element of this matrix is the similarity measure between two interest points. The new similarity measure is invariant to image rotation by taking into account the dominant orientation of the two interest points. The epipolar geometry constraint is finally imposed to reject the false matches. Experimental results on real images show this approach to be effective for general image matching.