A Comparison of SIFT, PCA-SIFT and SURF

This paper summarizes the three robust feature detection methods: ScaleInvariant Feature Transform (SIFT), Principal Component Analysis (PCA)–SIFTand Speeded Up Robust Features (SURF). This paper uses KNN (K-NearestNeighbor) and Random Sample Consensus (RANSAC) to the three methods inorder to analyze the results of the methods’ application in recognition. KNN isused to find the matches, and RANSAC to reject inconsistent matches fromwhich the inliers can take as correct matches. The performance of the robustfeature detection methods are compared for scale changes, rotation, blur,illumination changes and affine transformations. All the experiments userepeatability measurement and the number of correct matches for the evaluationmeasurements. SIFT presents its stability in most situations although it’s slow.SURF is the fastest one with good performance as the same as SIFT. PCA-SIFTshow its advantages in rotation and illumination changes.