基于广义形态滤波的图像去噪方法研究

针对现有去噪方法在噪声强度大时存在去噪效果差等不足,通过对形态学腐蚀和膨胀两个算子进行改进,提出了一种改进的广义形态滤波方法。该方法首先利用灰度值是否是3×3邻域的最大值或最小值来确定该像素是否被噪声污染,然后对被噪声污染的像素进行改进的广义形态滤波,而对没有被噪声污染的像素不进行处理。仿真结果表明,在噪声污染严重时提出的方法的峰值信噪比（PSNR）值比常用的中值滤波法可提高约4 dB以上。

Image-denoising Method based on Generalized Morphological Filter

For pear effect of the present denoising methods in strong-noise environment,and based on improvement of the morphological erosion operater and dilation operator,an improved image de-noising method based on generalized morphological filtering is proposed.Firstly,the polluted image pixel is determined by the gray value,which is the maximum or minimum value in the neighboring domain,3×3,and then the polluted pixel is handled by the improved generalized morphological filter.Simulation results show that this method has a 4 dB performance improvement as compared with the median filter when the image is severely polluted.