基于分水岭的提升小波图像去噪

Lifting Wavelet Image De-Noising Based on Watershed

为了图像去噪的同时能很好地保持图像的细节,提出了一种基于分水岭的提升小波图像去噪方法,先用分水岭分割方法检测出图像的分水岭脊线,提升小波去噪时就可用阈值去噪而不必担心损害图像的重要特征。其去噪步骤是:对噪声污染图像进行提升小波去噪;对原图像提取梯度幅度图像;对梯度图像平滑后进行分水岭变换;图像合并。实验结果表明,该方法不但可以保持图像的重要信息,而且能够提高去噪后图像的信噪比。

In order to preserve image details as well as canceling image noise, a lifting wavelet image denoising method based on watershed was presented. Watershed ridgeline of an image were detected with method of watershed firstly, and a threshold was used in lifting wavelet denoising without damaging image＇s characters. The first denosing process is lifting - wavelet denosing based on denosing-image;the second process is distilling gradsimage based on original- image;the third process is carrying watershed transform;the forth process is oombinating images. Experimental results showed that the algorithm could not only keep importance information of image, hut also could improve signal - to- noise ratio of an image.