Seam Carving in Image and Video

This book addresses the problem of retargeting, namely adapting large source images for effective viewing at a smaller size with possible applications to PDAs, or dynamic page layouts. Instead of extracting regions of interest for image retargeting, the uninteresting parts are removed from the scene. This is done by computing the RGB variance within non-overlapping 3x3 blocks and removing the block path with minimal variance cost using dynamic programming. It is shown that transformation to CIELAB space is more effective for visual interpretation of image content. The implementations are shown to be much faster than the seam carving approach. Schemes are also presented for speeding up the seam carving scheme itself. We have extended the approach to image expansion, object removal and object preservation. We also have worked on video retargeting using seam carving with additional constraint of time domain.

Autorentext

Awadhesh Srivastava is an Assistant Professor in IT Department of KIET Group of Institutions, Ghaziabad. He received M.Tech. in Computer Applications from IIT Delhi. He has guided many B.Tech. projects & having many publications in the area of Image and Video Processing. Human Gesture Recognition and E-learning are his areas of interest.