Title :
Saliency dectection using spatial features and spectral residual
Author :
Wang, Duyao ; Huang, Sujuan ; Wang, Shuozhong
Author_Institution :
Sch. of Commun. & Inf. Eng., Shanghai Univ., Shanghai, China
Abstract :
A saliency detection algorithm using spatial features and spectral residual is proposed in this paper. Feature maps including intensity, chromatic opponencies and motion are extracted from image and video, and pre-segmented. According to the segmentation results, saliency detection algorithm using spatial features and spectral residual is used to detect salient regions. Finally, the first-order entropy of the feature saliency map is used as weights to get the final saliency map. The proposed method takes global and local image contrast into account and achieves good results.
Keywords :
discrete Fourier transforms; entropy; feature extraction; image segmentation; object detection; video signal processing; discrete Fourier transform; feature extraction; feature saliency map; first-order entropy; global image contrast; image presegmentation; local image contrast; saliency detection; salient region; spatial features; spectral residual; video presegmentation; Conferences; Detection algorithms; Entropy; Feature extraction; Image segmentation; Motion segmentation; Visualization; DFT; Saliency detection; feature extraction; visual attention;
Conference_Titel :
Multimedia Technology (ICMT), 2011 International Conference on
Conference_Location :
Hangzhou
Print_ISBN :
978-1-61284-771-9
DOI :
10.1109/ICMT.2011.6002084