Automatically registering stereo pairs of tomographic imagery

A method for finding correspondences between points on a stero pair of 2-D electron micrographs of networks is developed. The objective for finding these matches is to combine stereo images of the same specimen to generate a 3-D representation. To perform the analysis, the gray level images of the micrographs are first binarized and thinned, and the network intersection points are detected. An algorithm which uses a geometric hashing scheme is then applied to detect corresponding points in the images. This algorithm is invariant to rotation, translation, and scale. The system has been tested on images of skeletal muscle tissue tubule networks.<
>