摘要
目前图像的去噪和边缘细节的保留是图像去噪中存在的2个大问题,该文提出使用边缘检测的方法检测出图像的边缘和纹理细节,对检测出的边缘和纹理细节图像进行处理后,将它和含噪图像分别进行小波变换,然后将其小波系数对应叠加,最后对叠加之后的小波系数进行小波自适应软阈值去噪。由于在软阈值去噪前叠加了边缘信息,因此边缘和细节部分得到了增强,虽然在阈值处理过程中由于边缘和细节均处于高频部分,在随后的软阈值去噪过程中存在被平滑的危险,但是增强后边缘和纹理的小波系数的幅值被放大,在阈值处理时可以得以保留。实验证明该方法比较wiener滤波在视觉效果和信噪比方面都有较大的改善,同时该方法比传统软阈值滤波,在视觉效果相差不大的情况下信噪比也有1~2个dB的提高。
In order to denoise and retain the edge and texture detail, this article proposes a method of image denoising based on adaptive soft-threshold and edge enhancement. To enhance edge, first it uses the edge detection to detect the edge of the image, and then preprocesses the edge image; next it can use wavelet transform to process the processed edge image and noise image, then it adds these wavelet coefficients at the corresponding site. So it can use the soft threshold denoising technology. Since the edge is enhanced before soft threshold, so the edge can be kept. From the result it can be found that comparing the wiener filter, this denoising method has good denoising image and the PSNR is higher. Comparing the soft threshold denoising, this method has 1~2 dB improvement in PSNR.
出处
《电子测量技术》
2008年第7期4-6,25,共4页
Electronic Measurement Technology
基金
航天支撑技术基金项目"利用图像再生技术提取细节的环保型图像恢复算法的研究"
中国地质大学优秀青年教师资助项目(CUGQNL0734)
关键词
边缘检测
软阈值去噪
边缘增强
edge detection
soft-threshold denoising
edge enhancement