Abstract:

The paper proposes a roubust solution for automatic matching of digital image pairs. The proposed solution combine two popular conventional approaches: area-based and feature-based image matching. In this research, characteristics of two conventional matching approaches are analyzing and a new matching solution is created. The area-based matching used as a preliminary step to eliminate gross matching errors. The RANSAC algorithm is used for exact matching two feature sets and projective model is used as space constraint of image pair. The reliability of proposed algorithm was validated on image pair taken from different cameras Sony Nex 5T, Cannon 550d.