自定聚焦对象的小景深图像效果模拟算法

摘要

Aiming at the disadvantage of ordinary digital camera that it can hardly get photos with nice small depth-of-field ( DOF) effect, we present a simulation algorithm which self-adjusts the image from large DOF to small DOF.In the algorithm, a local target area is first designated.On that basis, the target area segmentation is implemented through focusing object edge detection and region filling, then the segmented region is mended by close operation of mathematical morphology, in which the radius of structuring element is determined by the size of target area.The unfocused region is blurred at last, through this the fast small DOF simulation on original image is realised. Experiment proves that the method effectively improves the segmentation accuracy of focusing object, the applicability of the algorithm is high as well, and good simulation result of small DOF can be obtained for different target area.%针对普通数码相机难以拍摄出小景深特写效果的缺陷，提出一种由大景深图像自适应转换至小景深效果的模拟算法。算法在指定局部目标区域的基础上，通过聚焦对象边缘检测及区域填充实现目标区分割，并运用数学形态学闭运算对分割区进行修正，由目标区域的大小来确定结构元素半径，最后通过非聚焦区的模糊虚化实现原图的小景深快速模拟。实验证明，该方法有效提高了聚焦对象的分割精确度，且算法适应力强，对不同的目标区均能获得良好的小景深模拟效果。