A SABC-based Image Feature Point Matching Method

An Artificial Bee Colony Algorithm (ABC for short) based feature matching method is proposed in this article. The process for seeking for optimal honey source in the ABC algorithm is adopted to simulate feature matching. Four honey source generation constraints are proposed based on possible matching results; in addition, the epipolar geometry constraint is introduced to design for the formula of fitness. Besides, aiming at disadvantages of the conventional ABC algorithm such as the local optimum due to "early-maturing", the simulated annealing (SA for short) algorithm is introduced for improvement, referred to as SABC. Based on experimental results, the matching algorithm proposed in this article achieves favorable results on instantaneity and accuracy.