A stereo matching process for the recovery of depth information

Abstract

The authors present a stereo matching method that matches the Walsh coefficients for the recovery of the depth information in two-dimensional images. In this method, a stereo pair of images is transformed into the Walsh series and then decomposed into a pair of strings, carrying the first eight coefficients of the Walsh series as attributes. In order to explicitly integrate the uniqueness and the ordering constraints into the matching process, a string-to-string matching technique is applied to bring the two strings into correspondence. The experimental results obtained show that this method presents a very effective approach to the recovery of the depth information.<>