Video retargeting based on SH equalisation and seam carving

摘要

This study presents an efficient algorithm for video retargeting in the saliency histogram (SH) domain. It divides video frames into mesh quads based on the SH and reforms mesh grids at a quad level to fit the dimensions of a screen. Specifically, the first frame of a scene is retargeted by using the SH equalisation-seam carving (SHE-SC) algorithm, and the rest of the frames are classified into two types based on the content difference between consecutive frames, and then retargeted using different schemes. Experimental results show that the proposed algorithm is preferable to other existing algorithms.