The Algorithm Of Depth From Defocus With Gray Gradient Method

In computer vision, computing the distance between the goal object and the camera is very important. It is the key to 3D reconstructions. In recent years, the method of the depth from defocusing (DFD) is one of the hotpots in computer vision. It belongs to the monocular vision. It avoided characteristic points matching question in stereoscopic vision which has not been able to be solved effectively at present. Therefore it has practical application values in many fields. But the existing algorithm of DFD mostly uses the edge information of images through edge examination, and estimates the distance from object to the camera through the survey image edge fuzzy radius. However, the defocus image edge fuzzy degree is so big that it is difficult to determine its exact location. So these methods can not apply in many fields.This paper bases on the Subbarao's algorithm, and uses the gray gradient method in estimating the depth of defocus image. Firstly, determine the comparatively depth of the two relative points in the interface of the object according the gray gradient method. Secondly, judge the position relationship between actual image surface and focused image plane according to defocus model. Thirdly, make use of the S-transform to estimate the parameter of Point Spread Function, then obtains the formula that can estimate distance between the object and the lens.Through the fundamental research and analysis, the feasibility in using the gray information of a defocus image to estimate the depth of object is proved.