A stereo line matching technique for aerial images based on a pair-wise relation approach

Abstract

In this study, we developed a new pair-wise relation based approach for the matching of line features from stereo aerial images. To solve the final matching inconsistencies, we propose an iterative pair based post-processing algorithm in which the matching inconsistencies are eliminated using three novel measures and a final similarity voting scheme. The approach is tested over four urban test sites with various built-up characteristics, and for all test sites, we achieved a stereo line matching performance of 98%. The overall results indicate that the proposed approach is highly robust for the line features extracted in (very) dense urban areas.