Feature point detection and matching of wide baseline image based on scale space theory and guided matching algorithm

Image correspondence is a key issue in computer vision. Although many matching techniques in short baseline have been developed, the wide baseline correspondence problem with large scale, rotation, illumination and affine transformations is still not tackled very well The paper analyses the defect of the traditional method and proposes a new matching method which based on multi-scale Harris algorithm and two-way guided matching to achieve large number of accurate point correspondences between un-calibrated image sequences of the same scene for wide baseline. Experimental results show that the guided matching method can be used for severe scene variations and provide evidence of improved performance with respect to the SIFT distance and Harris matchers. It is also useful to the matching in short baseline, and the results of this method are better than that of the traditional method.