A multiresolution algorithm for aligning electrophoresis images

A multiresolution algorithm for geometry correction of electrophoretic gel images is presented. Because it uses a multiresolution image representation, the algorithm naturally incorporates the smoothness of the geometric distortion which such images typically display, but can cater with a much greater variability than simple landmark-based image registration. The algorithm has been tested on a wide variety of gel image data and has been found to be robust to variations in the image content which are typical of these images.