A computational vision approach to image registration

A computational vision approach is presented for the estimation of 2-D translation, rotation, and scale from two partially overlapping images. The approach results in a fast method that produces good results even when large rotation and translation have occurred between the two frames and the images are devoid of significant features. An illuminant direction estimation method is first used to obtain an initial estimate of camera rotation. A small number of feature points are then located based on a Gabor wavelet model for detecting local curvature discontinuities. An initial estimate of scale and translation is obtained by pairwise matching of the feature points detected from both frames. Finally, hierarchical feature matching is performed to obtain an accurate estimate of translation, rotation and scale. A method for error analysis on matching results is also presented. Experiments with synthetic and real images show that this algorithm yields accurate results when the scale of the images differs by up to 10%, the overlap between the two frames is as small as 23%, and the camera rotation between the two frames is significant. Experimental results on several Mojave desert images acquired from a balloon are presented. Applications of the method to texture and stereo image registration, and satellite image mosaicking are presented.