Detection of the most interesting region in an image has become an important subject in computer vision. Bottom-up model detects regions that differ with respect to their surrounding ones. These regions are known as salient regions. In this paper, we propose a new bottom-up model for saliency detection using the color of background regions. In the model, first, the image is segmented into superpixels. The boundary superpixels of the image are considered as background and others as uncertain superpixels. Then, the saliency is determined based on color difference between each uncertain superpixel and all background superpixels in the CIE LAB space. The proposed method can highlight the whole object regions uniformly and suppress the background regions effectively. Experimental results on the MSRA-1000 dataset demonstrate our method performs well against the state-of-the-art methods in terms of speed and accuracy.