Salient object detection is one of the most challenging areas in computer vision and has extensive applications in many fields. In this paper, two novel features, such as manifold ranking-based compactness and foreground connectivity, are designed in the proposed model. The new designed compactness is constructed by integrating two compactness maps which are, respectively, weighted by the spatial and central contrast of target region to all regions in the image. The foreground connectivity is obtained based on the novel compactness and geodesic distance. Since multiscale salient detections highlight different parts of the objects, we fuse four saliency maps on different scales to further improve the performance of the detection. Experiments on three public benchmark datasets demonstrate that the proposed method improves the accuracy of saliency detection and performs better than 14 state-of-the-art methods.