摘要
利用稀疏表示的自适应特征,将稀疏表示的多分辨理论应用于图像的去噪处理中,提出了一种基于稀疏表示的图像分块去噪方法。首先将噪声图像分割成一定尺寸的图像块,选出同质块与非同质块;然后利用小波去噪方法处理同质块,而采用脊波去噪方法处理非同质块,从而得到去噪后的图像;最后采用维纳滤波器对去噪后的图像进一步处理。实验结果表明,该方法与单纯的小波去噪方法和脊波去噪方法相比,信噪比有了较高的改善,较好地去除图像噪声,并且很好地保存图像的边缘纹理信息。
Adaptive features of the sparse representation , the sparse representation theory is applied to multiresolution image denoising proposed method based sparse representation of the image block .First, the noise image is divided into image blocks of a certain size , selected homogenous block and a non-ho-mogenous block;then wavelet denoising processing homogenous block , while the use ridgelet denoising method to deal with non-homogenous block , thereby obtaining image after go noise .Finally using Wiener filter further process denoising image .Experimental results show that this method compared with pure wavelet denoising and ridgelet denoising method , signal-to-noise ratio has been improved high , remove the image noise , and to save the edge of the image texture information .
出处
《工业仪表与自动化装置》
2013年第5期13-16,共4页
Industrial Instrumentation & Automation
基金
黑龙江省教育厅科学技术研究项目资助(12533054)
关键词
图像去噪
稀疏表示
小波变换
脊波变换
image denoising
sparse representation
wavelet transform
ridgelet transform
