基于图像信噪比选择优化高斯滤波尺度

该文提出了一种选择优化高斯滤波尺度的算法。首先对图像的信噪比进行估计，以此来评估图像受噪声退化的程度；然后在样本图像中加上不同程度的噪声，利用最优化准则得到相应的最优高斯滤波尺度；最后在统计意义上研究最优化高斯尺度和图像信噪比之间的关系，通过数据拟合，可得到基于图像信噪比来选择高斯滤波尺度的算法。在Canny边界探测和非线性扩散滤波上的应用实验，显示了新算法的有效性。

Selecting the Optimal Gaussian Filtering Scale via the SNR of Image

This paper presents an algorithm to select the optimal Gaussian filtering scale. First, it estimates the SNR of image to evaluate the noise degraded intensity of image. Second, it obtains the optimal Gaussian filtering scales by using optimal criteria on the sample images with varying intensity of noise. Finally, a statistically investigation of the relationship between the optimal Gaussian scales and their corresponding SNR and using the method of data fitting lead to the algorithm of estimating Gaussian filtering scale based on the SNR of image. Applied experiments on the algorithms of Canny edge detection and the nonlinear diffusion filtering demonstrate the efficiency of the proposed algorithm.