A fast image matching method based on high-dimensional combined features

In recent years, image matching navigation technology has been developing rapidly, but it is hard to meet the actual requirement of real-time. In this paper, a fast image matching method based on high-dimensional combined features is proposed, and K-dimensional tree (KD-tree) and particle swarm optimization (PSO) algorithm are introduced to improve the matching speed. At first we present a new concept of high-dimensional combined feature, and construct the features of two adjacent frames in sequence images as matching primitives. Then the position relation of two adjacent frame images can be determined according to the geometric constraints among the features. Finally, we introduce KD-tree and PSO algorithm to optimize the search process. The simulation results show that the matching is still completed at the rotation angle of −5 ° to 5 ° and the scale factor of 0.9 to 1.1, meanwhile, the time consumption is within 1 second. As a conclusion, the algorithm can effectively improve the real-time performance of image matching, and is robust to rotation and scale changes, which satisfies the requirements of navigation system.