Image Registration of Multi-View Satellite Images Using Best Feature Points Detection and Matching Methods from SURF, SIFT and PCA-SIFT
Utsav Shah, Darsshana Mistry, Asim Banerjee · 2014
Image Registration (IR) is a process of arranging two images (references and sense images) of the same scene taken at different times, from different sensors, and/or different viewpoints into a common coordinate system. There are four basic steps of image registration procedures: feature detection, feature matching, and transform model estimation and image transformation and re-sampling. In Multi-view analysis images of the same scene are acquired from different viewpoints. The aim behind this methodology is to gain larger a 2D view representation of the scanned scene. Features can be found by detection of point, interest points, corners, edges, lines, blobs, T-junctions etc. This paper summarizes the three robust feature detection and matching methods: Scale Invariant Feature Transform (SIFT), Principal Component Analysis (PCA)–SIFT and Speeded Up Robust Features (SURF). SIFT find its interest points using Difference of Gaussian (DoG). SIFT presents its stability in most situations although it’s slow. PCA-SIFT show its advantages in rotation and illumination change and it is faster than SIFT. SURF is the fastest one with good performance as the same as SIFT. ‘Fast-Hessian’ detector that used in SURF is more than 3 times faster that DOG