Image of Ship Target Matching Algorithm Based on PCA-SIFT in the Complicated Environments with Noise

Abstract

In the complicated environments with noise, the traditional SIFT algorithm has the low matching speed of extracting feature points with the weak real-time performance, this paper proposes a kind of feature point matching algorithm for ship images based on principal component analysis and Scale-Invariant Feature Transform (PCA-SIFT) features. Firstly, the source images are added with the Gauss noise, impulse noise and multiplicative noise to generated the images to be matched; then we extract the feature vector from two images and use principal component analysis to reduce the dimension of feature vectors; the nearest neighbor algorithm to match feature points. The random sample consensus (RANSAC) algorithm can be used to eliminate the error matching. The experimental results show that, in the ship image, the PCA-SIFT algorithm can effectively reduce the number of matching feature points and description vector dimension, accelerate the matching speed and achieve high precision with the good stability, especially in the complicated environments with noise.