Skip to Main Content
Why is measuring image similarity useful? There are abundant computer imaging applications requiring some kind of similarity measurement as part of their processes. Although the applications are quite varied, and the implementation details of each solution are unique, all share the common thread in that features or attributes of the image (in each specific application) are measured and then compared to other features from a database of images or with some reference model to extract some meaningful conclusions or functionality about the image data on hand. This paper describes several methods of measuring image similarity: a pattern recognition approach, comparison of frames in a video sequence, image stabilization using a homographic transformation, and using image feature points to compute similarities and generate an image mosaic.