Region Based Non-Uniform Scheme for Effective Retargeting

Abstract

With the invent of displays of varied features like bendable displays, twistable displays and slant displays the feed of these displays need to be adjusted so that useful objects of the scene are not destroyed. In general, if we want to project a particular scene of specified size after capturing on the display of other dimensions, image resizing is done. In image resizing, the whole image will be converted into dimension of the target display. In the process the whole image, including all the objects may be expanded or squeezed depending on the size of the target display. But, in image retargeting, in addition to changing the size of the image to suit the target display, some additional intelligence will be added there by preserving or retaining useful objects and removing unnecessary background information. In this paper, the global significance map will drive the retargeting which is formed by merging the local significance maps. The local significance maps are derived after forming regions by considering the average intensity levels locally. The simulation results clearly show the superiority of the scheme proposed