A real-time image mosaicing using FAST detector and FREAK descriptor

Abstract

Image mosaicing is a technique of combining information of different images of the same scene to obtain more information in a single snap that contains every image data. In this paper a real-time image mosaicing is implemented by capturing multiple images of a continues scene with slight variation in the Field of View (FoV) of a camera sensor. Features from Accelerated Segment Test (FAST) feature detector detects invariant robust feature points of each image and Fast Retina Key point (FREAK) binary descriptor extracts feature vectors for each and every feature point. Feature matching is executed with the computation of the hamming distance between the feature points, hamming distance is high for more discriminative feature elements. Mapping of feature points is achieved with Homography computation. Distortions of an image can be corrected with image warping process. Robust corner points and corner point vectors are calculated for each input image followed by the number of matching points of corresponding consecutive images validate the effectiveness of FAST and FREAK algorithms. A comparison of traditional algorithms like Scale Invariant Features Transform (SIFT), Speed Up Robust Features (SURF) shows that our experimental results with FAST and FREAK are more accurate, faster and robust.