一种基于人眼视觉注意力的三维视频重定向方法

3D video retargeting based on human visual attention

从人眼视觉注意力的角度出发,提出了一种三维视频重定位方法。首先,引入场景的深度信息,并结合梯度、显著性和运动对象检测构造模型,得到视觉注意力能量项;然后,分别构造帧内能量项和帧间能量项,并结合视觉注意力能量项,构造总能量函数进行细缝裁剪,实现视频的重定向。实验结果表明,本文方法能够较好地保护视频场景中的重要对象,并避免较大的失真和抖动。

When the resolution of the image/video is not consistent with that of the display devices,the problem of adaptively matching screen resolution is known as retargeting.Recently,content-aware retargeting approaches have attracted more attention from the perspective of user′s quality of experience.In this paper,a 3Dvideo retargeting method based on human visual attention is proposed.First,we model the visual attention energy by combining gradient,saliency and motion object information.Then,we construct intra and inter frame energy terms respectively,and obtain the total energy for seam carving by combining intra frame energy,inter frame energy and visual attention energy.Experimental results show that the proposed video retargeting method can protect the important objects in video very well without serious distortion or jitter.