科研之友
微信
新浪微博
Facebook
分享链接
摘要
Video services achieved from the cloud computing become a critical concern. In various image processing applications, the organizations are slow in accepting it due to security issues and challenges associated with it. In terms of security, cloud-based video services must be managed and operated at equivalent levels to enterprise systems. For various image processing applications, saliency detection can be used to extract the regions of interest for images. Compared with images, the motion feature has to be considered for video saliency detection. This paper proposes a new video saliency detection model based on human visual acuity and spatiotemporal cues in cloud systems. Firstly, each video frame is divided into small image patches. Four features (including one intensity, one motion and two colour features) are extracted from each image patch. Based on these four features, the Quaternion Fourier Transform is used to obtain the amplitude spectrum for each image patch. The saliency value of each image patch is determined by two factors: the amplitude difference between this image patch and other image patch in the video frame, and the corresponding weighting of visual impact by human visual acuity. Experimental results demonstrate that the performance of the proposed video saliency detection model in cloud systems is better than other existing ones.
