Research on 3D Shape Reconstruction using Uneven Defocusing Model

This paper proposes a new method to reconstruct 3D shape from uneven defocusing image. It supposes that the defocusing model is not uniform. This method is also based on depth from defocus(DFD), there have been some methods in restoring the depth from the uniform defocusing image , but if we want to gain the 3D shape, we can only adopt the uneven defocusing model. This paper try to solve this problem in a new method based on such model. With the improved counter-filter similar to the wiener filter, we can obtain an image sequence through changing the value of the parameter which has close relationship with the radius of the blur circle. Then, we calculate the level of focusing and defocusing of each point in the image sequence to look for the most focusing point. We define the parameter of the most focusing point as the blurring circle of the point in the uneven defocusing image. Then, we use gradient operators as the clarity-evaluation operator. With the establishment of relationship between the defocus measurement and the depth transformation of the image sequence, we obtain the depth information of the object. As a test of validity of this method, we performed an experiment to process an uneven defocusing image and the experiment results showed that the method is valid for 3D reconstruction from uneven defocusing image.