Abstract: This paper proposes a new method to estimate the parameter of Point Spread Function（PSF） for out-of-focus Image.At first，the out-of-focus blurring is modeled as Gaussian blurring.Then，analyze the differential features of several types of image edges and extract Gaussian Line Spread Function（LSF） from image edges using these differential features.Since matching filtering has the advantage of maximizing the output signal-to-noise ratio，based on this philosophy this paper applies template matching method to the evaluation of LSF standard deviation（STDEV）.The STDEV of Gaussian PSF can be easily calculated from the STDEV of LSF by the symmetric property of PSF.The result shows that the proposed method can accurately evaluate the STDEV of out-of-focus image PSF.Finally，this paper takes the STDEV of PSF as an objective metric of image blurriness.The testing result shows that this objective metric correlates well with the judgment of human observers.Therefore，the STDEV of PSF is an excellent quality metric for out-of-focus blur images.

Key words: out-of-focus blur, point spread function, line spread function, image blurriness, template matching

摘要： 提出一种新的散焦图像模糊度估计方法。由图像中的阶跃型边缘提取线扩展函数。根据匹配滤波可实现最优信噪比信号检测的思想，引入模板匹配的方法计算线扩展函数的标准差。由点扩展函数的圆周对称性从线扩展函数的标准差得到高斯型点扩展函数的标准差。实验结果表明，该方法能够准确地估计散焦模糊图像点扩展函数的标准差。将点扩展函数的标准差作为一种评价图像模糊度的测度。实验表明该测度符合人眼的视觉特性，可以很好地判定散焦模糊图像的模糊度。

关键词: 散焦模糊, 点扩展函数, 线扩展函数, 图像模糊度, 模板匹配