Title :
Saliency detection on videos with scene change
Author :
Junling Li ; Fang Meng ; Jingbo Mao
Author_Institution :
Inf. Eng. Sch., Commun. Univ. of China, Beijing, China
Abstract :
In this paper, we propose an improved framework of video saliency detection, which is especially suitable for video sequences with scene changes, a challenge for previously algorithms used spatiotemporal saliency detection. We present the scene change detection mechanism as an essential module of our framework to adjust temporal parameters in video saliency detection. The paper analyzes the necessity of scene change detection mechanism firstly. After that, we propose a simple but robust method for detecting scene change and illustrate its working mechanism in our framework. Experimental results demonstrate that our algorithm outperforms the corresponding original one by improving the accuracy of saliency greatly.
Keywords :
video signal processing; scene change detection mechanism; spatiotemporal saliency detection; video saliency detection; video sequences; Accuracy; Feature extraction; Interference; Spatiotemporal phenomena; Switches; Videos; Visualization; bottom-up saliency; inter frame correlation; saptial-temporal saliency; scene change; video saliency;
Conference_Titel :
Audio, Language and Image Processing (ICALIP), 2014 International Conference on
Conference_Location :
Shanghai
Print_ISBN :
978-1-4799-3902-2
DOI :
10.1109/ICALIP.2014.7009845