Simultaneously recovering affine motion and defocus blur using moments

In practical computer vision system, there are situations where displacement of an image is accompanied by a defocus blur. Since depth can be estimated from the degree of blur, in this paper we propose a method to get the blur parameter and simultaneously affine transformation parameters. Firstly, we introduce a method to get the motion parameters from the original image and the affine transformed and simultaneously blurred image(second image). Then we use these parameters to construct affine transformed image. The last step is to get the blur parameters from the constructed image and the second image. The effectiveness of the proposed method is demonstrated by experimental results.