An Image Matching Algorithm Based on SIFT and Improved LTP

Abstract

SIFT is one of the most robust and widely used image matching algorithms based on local features. But the key-points descriptor of SIFT algorithm have 128 dimensions. Aiming to the problem of its high dimension and complexity, a novel image matching algorithm is proposed. The descriptors of SIFT key-points are constructed by the rotation invariant LTP, city-block distance is also employed to reduce calculation of key-points matching. The experiment is achieved through different lighting, blur changes and rotation of images, the results show that this method can reduce the processing time and raise image matching efficiency.