We propose an object-level salient detection algorithm which explicitly explores bottom-up visual attention and objectness cues.
Some category-independent object candidates are firstly segmented from the image by the quantized color attributes of images.
Global cues and candidate objectness are developed, to evaluate bottom-up visual attention of the whole image and the objectness of the object candidates respectively.
By explicitly combining local objectness cue with global low-level saliency cues with candidates location and color attributes, our proposed method is more suitable for processing images with complex background.