Method for Depth Map Extraction of Defocused Image Based on Segmentation

For a large number of defocused images in films and televisions,the paper proposes a method for depth map extraction of defocused image.Firstly,the focused foreground is separated from the defocused background in a defocused image.For the defocused background,a method for depth map model matching is presented,a depth map model is constructed,and the background and the corresponding depth map model is matched with the help of human's sensitive judgments on the scene depth.The paper also presents the depth map reprocessing based on color segmentation to improve the accuracy of the depth map of the scene.Finally,for the focused foreground,we assign a single value to be the depth in the area of the foreground,and produce the final depth map with respect to background depth map.The experiment shows depth maps obtained by the proposed method have good results in the areas of depth smooth and depth jump.