Feature matching algorithm based on KAZE and fast approximate nearest neighbor search
Cai Ze-Ping, Degui Xiao · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2014
This paper proposed a feature matching algorithm based on KAZE and fast approximate nearest neighbor search for that SIFT and SURF feature detection algorithm,extracting features by Gaussian pyramid in the linear scale space has the problems of fuzzy boundaries, detail missing, and low feature points matching rate.First, this algorithm uses Additive Operator Splitting(AOS) method for nonlinear diffusion filtering,and structure nonlinear scale space.Then use Hessian matrix to detect feature points, and construct the Gauge-SURF(G-SURF) descriptions in the Gauge coordinate.Finally adopt fast approximate nearest neighbor search algorithm to match feature points, and use RANSAC algorithm to eliminate false matching points.Experiments show that this algorithm ensures the real-time nature and improves the feature matching rate.