Fast image matching based-on improved SURF algorithm

Abstract

In this paper, to solve the problems that matching an image through the SURF (Speeded Up Robust Features) algorithm spends too much time, a novel image matching algorithm to improve surf will be introduced. While surf is known to be strong but computationally still expensive, it has not attained real-time performance yet. Thus this algorithm is based on the heavily modified SURF feature descriptors to attain the faster speed. We modify the Fast-Hessian Detector of SURF for the FAST (Features from Accelerated Segment Test) corner detector to match an image. This improved surf in this paper to match an image can fulfill the real-time requirement and get a considerable accurate result on computers.