In order to recognize the salient objects in images more accurately, a saliency detection algorithm based on foreground optimization and probability estimation is proposed. The proposed algorithm mainly includes three parts, i. e., foreground and background cues selection, foreground cues optimization and salient region detection based on probability estimation. Firstly, the simple linear iterative clustering algorithm is used for initial segmentation of images. Then, the background cues and foreground cues are detected from the given image respectively. In addition, the foreground cues are optimized by background cues. Finally, saliency maps are obtained using the probability estimation algorithm on the basis of the background cues and optimized foreground cues respectively. The results are fused to generate final saliency map. Experimental results show that the proposed algorithm achieves higher precision in comparison with other algorithms, which has better detection performance.