Fast Scale-Invariant Feature Transform using Niching Particle Swarm Optimization

T. Jeon and M. Jeon (Korea)

Keywords

Image registration, scaleinvariant feature transform, niching particle swarm optimization, difference of Gaussian,scalespace extrema detection.

Abstract

This paper presents a fast scale-invariant feature transform (SIFT) using a niching particle swarm optimization based on similarity between Difference-of-Gaussian (DoG) images. The aim of proposed method is to detect extrema in scale-space with converting exhaustive search problem into multiple solutions optimization problem. The niching particle swarm optimization model with modified initialization step finds effectively local extrema based on similarity of adjacent DoG images. Experimental results show that the proposed model outperforms previous SIFT algorithms in computational complexity and the number of detected features.

Important Links:



Go Back