IMPROVING SIFT FOR IMAGE FEATURE EXTRACTION

Abstract

This paper reviews a classical image feature extraction algorithm, namely SIFT (i.e. Scale Invariant Feature Transform) and modifies it in order to increase its repeatability score. We are using an approach that is inspired from another computer vision algorithm, namely FAST. The tests presented in the evaluation section show that our approach (i.e. SIFT-FAST) obtains better repeatability scores over classical SIFT.

Keywords

• image feature extraction, SIFT, FAST.