Application of an adaptive least squares correlation algorithm for stereo matching planetary image data

Abstract

An adaptive least squares (ALS) correlation algorithm and a region-growing algorithm were implemented into a stereo-matching computer program. This program (referred to as the Stereo Matching Tool Kit (SMTK)) was designed specifically for the application to planetary image data. The ALS algorithm matches a patch of one image to the corresponding area in a second image. The matching procedure is an iterative process that minimizes the sum of the square differences between the two patches to determine an optimal set of transformation parameters. Successful matches are then used to predict potential match points for surrounding locations. Using potential match points in conjunction with the region-growing algorithm, a population of match points between the two images is determined. The stereo-matching process is initiated by using the ALS algorithm in conjunction with Spacecraft, Planet, Instrument, C-matrix, and Events (SPICE) information to automatically determine a set of seed points. SMTK was tested on two planetary image data sets: Mariner 10 and Clementine. SMTK-derived digital elevation models compare well with topography generated by an area-based stereo matcher requiting manual selection of seed points, analog stereo techniques, and photoclinometry. Copyright 2008 by the American Geophysical Union.