A Novel Approach to Seamless Image Stitching using S-FREAK and RANSAC Algorithm
Gopika Krishnan G, Dr.Priya S, Ashok Kumar T · Journal of Critical Reviews · 2020
Image stitching is increasingly becoming popular in the fields of image processing, computer vision, virtual reality, computer graphics, human computer interaction and multimedia. Image stitching or mosaicing is a technique in which several images of overlapping domain of view are combined together for a panoramic image of high resolution. Image stitching surveys depict that it is still now a puzzling problem for the construction of panoramic images. The input to a stitching algorithm is multiple, overlapping images captured from different camera views and the output is a panorama of wider field of view made by merging and stitching the individual images. Feature extraction, feature matching, homography estimation and stitching are the steps performed to make panoramic images. Traditional image stitching methods based on different feature extraction techniques require long registration time for high resolution images. In this work, a novel image stitching method based on S-FREAK is proposed by combining SIFT (Scale Invariant Feature Transform) and FREAK (Fast Retina Keypoints). The feature descriptors from one image are matched with the other to find the best closeness and only the features with best closeness are kept while the rest ones are discarded. A transformation model is estimated from these features and the image is warped correspondingly. Image stitching is a technology for solving the field of view (FOV) limitation in images.