Panoramic view creation using invariant momentsand SURF features

Abstract

This paper presents an efficient approach for panoramic view creation to enlarge the field of view. Image stitching is used to integrate information from multiple images with overlapping fields of view in order to produce a panoramic view with all the contents fitted into a single frame. The major contribution of the proposed work resides in the preliminary selection of overlapping region using gradient and invariant moments approach. This step plays a vital role as it eliminates the need to extract features from the unnecessary portions of the input image. Once the overlapping regionsare identified, SURF(Speeded Up Robust Features) features are extracted from it. These features are matched for finding the transformation matrix using RANSAC (Random Sample Consensus) algorithm. The mosaic is obtained by warping the images into a single frame. This stitching method is iteratively applied to a series of images to create a complete panoramic view. In order to justify the effectiveness of the proposed work, it is compared against traditional methods in terms of computational complexity for feature extraction.