局部高斯尺度混合模型的傅里叶-小波图像降噪  

Hybrid Fourier-wavelet image denoising using local Gaussian scale mixtures model

作　　者：唐锐[1] 张敬东[2] 张祺[1] 

机构地区：[1]攀枝花学院机电工程学院,四川攀枝花617000 [2]攀枝花学院交通与汽车工程学院,四川攀枝花617000

出　　处：《激光与红外》2013年第5期592-595,共4页Laser & Infrared

基　　金：四川省教育厅自然科学重点项目(No.11ZA253)资助

摘　　要：提出了一种基于局部高斯尺度混合统计模型的傅里叶-小波图像降噪方法。所提出的降噪方法综合了两者的优点,考虑到噪声小波系数间的相关性,小波系数统计特性通过局部高斯尺度混合统计模型来刻画。实验结果表明,此法可有效去除噪声,并且能够克服传统的小波去噪效果与选用的小波基函数相关的局限性,和其他方法相比,无论从视觉上还是峰值信噪比上比较,此方法降噪效果明显较好。A hybrid Fourier-wavelet image de-noising using Local Gaussian Scale Mixtures Model is proposed.The de-noising method has overcome the disadvantage of Fourier transform and wavelet transform.It takes into account the correlation between wavelet coefficients of noise.The Local Gaussian Scale Mixtures Model is used to estimate the wavelet transform coefficients.Experiment results reveals that the proposed method can perform a good performance on de-noising.As the de-noise result of traditional wavelet is related to the wavelet basis function,this method overcome the limitation.Through experiments with images contaminated by additive random noise,we demonstrate that the performance of this method substantially surpasses the previously published methods,both visually and in terms of signal to noise ratio.

关 键 词：图像降噪 傅里叶变换 局部高斯尺度混合模型 小波变换 峰值信噪比 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]