基于Radon变换估计点扩散函数

Based on the Radon transform to estimate the Point Spread Function

目前通过点扩散函数(PSF)来恢复高清图像都假设了点扩散函数的某些性能是已知的,比如类型,尺寸等。本文通过利用模糊图像边缘信息得到点扩散函数最大直径,然后利用Radon变换法估计点扩散函数。本文是基于完全未知的点扩散函数来逐步得到其相关特性,而且通过大量实验证明由于噪声的影响,直接利用Radon变换估计点扩散函数只能得到点扩散函数的大致结构,恢复出来的图像效果欠佳,所以需要先对原始模糊图像进行滤波,并对得到的点扩散函数框架做插值运算,可以得到更准确的点扩散函数。

Currently, using the point spread function (PSF) to get the deblurring images also needs to assume that some properties of the PSF are known,for example, type, size, etc. This paper uses the edges of blurred images to estimate the maximum diameter of the PSF, then uses the Radon transform to obtain the PSF. This paper is based on the completely un-known PSF to gradually get its relevant characteristics. While more expensive, because of the noise, using the Radon trans-form directly to estimate the PSF only can get the general structure of the PSF, the quality of recovered images is poor. so the blurred images should be filtered, and the general structure of the PSF should be interpolated, the PSF will perform better.