基于分割的离焦图像深度图提取方法  

Method for Depth Map Extraction of Defocused Image Based on Segmentation

作　　者：史国凯[1] 王琼华[1,2] 李大海[1] 赵悟翔[1] 彭华荣[1] 罗江勇[1] 

机构地区：[1]四川大学电子信息学院,四川成都610065 [2]四川大学视觉合成图形图像技术重点学科实验室,四川成都610065

出　　处：《液晶与显示》2012年第2期229-234,共6页Chinese Journal of Liquid Crystals and Displays

基　　金：国家自然科学基金(No.60877004;No.61036008)

摘　　要：针对影视作品中的大量离焦图像,提出了一种离焦图像的深度图提取方法。将离焦图像的聚焦前景和离焦背景进行分离。对离焦背景提出了深度图模型匹配的方法,构建深度图模型并结合人眼视觉对场景深度的敏锐判断,将背景与对应的深度图模型进行匹配,实现背景深度图的构建;提出了基于颜色分割的深度图再处理,来进一步提高场景深度图的精度。对前景采用单深度赋值,并结合背景深度图融合生成最终深度图。实验表明采用该方法提取的深度图在深度跳跃和深度平滑区域都得到了好的效果。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.

关 键 词：2D转3D 前景分割 深度图 颜色分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]