Saliency estimation is a complex problem of computer vision
area, which results can enhance many tools or applications. In most of existing solutions, data that comes from stereopsis was not involved. We propose an algorithm, that adds depth information to detection of important
objects in the scene. We show and confirm, that the whole problem of salient objects detection can be decomposed into series of simple image processing operations. To verify functioning and performance of the method, we test it on available RGBD datasets.