Visual Preserving Video Retargeting With Deformable Shape Consistency

With the development of different kinds of mobile devices andInternet technology, more and more people choose to view images andvideos on the devices like cellphones, PDAs, tablet PCs. However, weneed to retarget image and video to meet the resolution and sizerequirements of different equipment. Therefore, Content-based image andvideo retargeting method becomes a hot and difficult point for the presentstudy. Video retargeting is a technique that resizes the aspect ratio andresolution of a video in a content aware manner. There are three mainchallenges in the video retargeting: visual saliency preservation,deformation prevention and temporal consistency preservation. Existingvideo retargeting methods can be roughly classified into seam carvingbased methods and warping based methods. These methods effectivelyprotect important regions in video, but can hardly deal with deformationsand temporal incoherence, especially for videos whose foreground andbackground are both in motion.In this paper, we propose a novel shape preserving video retargetingmethod to reduce deformations and maintain temporal consistency viamatching salient curves in the video frames. In order to solve shapepreserving, our algorithm use Our method transforms the anti-deformationand temporal consistency problems in video retargeting into a curvematching cost minimization problem. By incorporating a deformation costand temporal inconsistency cost into the seam carving framework, thequality of the retargeted videos can be significantly improved.Experimental results also show that the proposed method achieves betterperformance compared with other two state-of-the-art methods.