关键词: 图像恢复 散焦模糊图像 点扩散函数 鉴别 拉普拉斯算子

Abstract:

The point spread function（PSF） of the defocus image is approximately columniform. It is reasonable to equate identifying the PSF with identifying th e radius of the column. A new method of identifying the radius is given. There are three important computing steps. The first step is to derivate the de  focus image with a Laplacian operator, which is an isotropic second difference arithmetic operator. The second step is to calculate the autocorrelation   of the derivative image. There must be an annular groove in the 3D show of the autocorrelation image. The bottom of the annular groove is linked by the  autocorrelation peaks whose values are minus. The positions of the peaks link a cycle, whose radius is double that of the PSF column＇s. The third step   is to transform the X-Y coordinates into polar coordinates for the autocorrelation by the interpolation. An identifyication curve will be gained by integral calculus of the angle. The radius of the PSF column can be identified directly from the curve with high precision. This new method has the power t  o resist noise and does not cost much computing time.

Key words: （image recovery;defocus image, point spread function, identification, Laplacian operator）