Abstract

Ukendra Kumar Sahu, Sachin Meshram

ABSTRACT

Traditional methods used to produce dense disparity maps for given stereo image pairs is classified as either area based or feature based methods. In this paper, we make use of area-based approach for selection of control points which are further used in order to calculate matching pixel using the surface fitting approach that performs pixel matching in stereo-pairs. The approach performs pixel matching in conjugate pairs in two stages. The 1st stage of similarity between windows is measured using Normalized Cross Correlation if maximum Normalized Cross Correlation calculated exceeds a certain upper threshold it is considered as the match and the corresponding pixels are selected as control points and are stored in the database. In the 2nd stage of the algorithm, surface fitting is performed in this a window is selected such that at least 7 control points are obtained in case number of control points obtained are less window size increases dynamically so that more control points can be incorporated. With the multistage approach & the parallel processing speed of dense disparity map generation will improve along with improved accuracy. The method is tolerant to radiometric distortions and occlusions. INDEX TERMS—