几种图像去噪方法的比较研究

Comparison and Research on Several Image Denoising Methods

针对目前在传输过程中图像会被多种噪声影响和干扰的问题,利用MATLAB软件平台,通过对图像去噪后的定性、定量分析,研究和比较了几种图像去噪方法。实验结果表明,对于含椒盐噪声的图像,中值去噪方法效果较好;对于含高斯噪声的图像,空域低通滤波(高斯模板)去噪方法较好;对于含散斑噪声的图像,小波去噪方法效果较好。整体可以看出,对于三种噪声,小波变换的图像去噪方法因其良好的时频特性和多分辨特性,去噪的效果相对其他去噪方法较稳定,适合于多种噪声,使得小波去噪方法比传统算法具有更好的去噪效果。

Aiming at the problem that the image would be affected and disturbed by many kinds of noises in the process of transmission, with MATLAB software platform, the qualitative and quantitative analyses of image denoising are done, and several image denoising methods are discussed and compared. The experimental results indicate that the median denoising method is better for images containing salt and pepper noise, while the spatial low-pass filtering （Gauss template） denoising method better for the image with Gauss noise, and the wavelet denoising method better for speckle noise images. It can be seen that for the whole three kinds of noise, the image denoising method of wavelet transform is more stable than other denoising methods for its good time-frequency and multi-resolution characteristics. Wavelet denoising method is suitable for a variety of noise, and has better denoising effect than the traditional denoising algorithms.