The Study of Preprocessing and Postprocessing Techniques of Image Stitching

摘要

Image stitching played a vital role in development of real life applications in Computer Vision field. The idea behind Image Stitching is to construct seamless high definition panoramic view using different viewpoints with multiple smaller overlapping images. Image stitching method involves steps such as feature detection and description, feature matching and homography estimation which are referred as preprocessing techniques. After this, postprocessing techniques like finalization of reference frame, image blending and compositing methods are studied. In this paper, the comparative analysis of above techniques is addressed quantitatively and qualitatively. Based on above, general ideas and selection of appropriate algorithms are provided to obtain landscape panorama.