摘要
在图像处理中,基于离散小波变换的提升算法比传统的卷积算法运算简单、实时性好、易于实现,因而被新的图像去噪所采用。小波提升算法是一种新的双正交小波构造方法,通过预测算子,确定高频信息,并初步确定低频信息,然后通过更新算子,对初步确定的低频信息进行修正,从而确定低频信息。笔者以该提升方法为基础,通过对初始D4双正交滤波器组进行提升和对偶提升,来获得不同的提升算子和对偶提升算子,从而构造出具有理想特性的新小波。通过实验数据和分析表明:笔者提出的算法和软阈值法结合比其他小波更能有效地取出图像噪声。
In image processing, the lifting scheme based on discrete wavelet transform is simpler in operation than the traditional convolution algorithm. For its good real - time characteristics and easy realization, it has been adopted by new image de - noise. The lifting scheme and realization algorithms of D4 wavelets in image de - noise are introduced. This method is based on lifting scheme. The initial set of finite biorthogonal filters is followed by lifting and dual lifting, and different lifting operator and dual lifting operator are obtained to construct new wavelet with particular properties. The experimental results and data analysis show that it is effective for the lifting scheme with time soft thresholding algorithm in de -noise process, and it is better than other wavelet methods.
出处
《武汉理工大学学报（信息与管理工程版）》
CAS
2006年第11期13-15,19,共4页
Journal of Wuhan University of Technology：Information & Management Engineering
基金
北京大学视觉与听觉信息处理国家实验室基金资助项目(2003005)
关键词
小波变换
提升方案
软阈值法
图像去噪
wavelet transform
lifting scheme
thresholding algorithm
image de - noise
