摘要
针对小波阈值去噪中的VisuShrink阈值(统一阈值)“过扼杀”细节系数和SUREShrink阈值(Stein无偏估计阈值)“过保留”噪声系数的特点,提出了一种新的模糊小波阈值去噪方法。该方法根据小波系数局部方差能体现信号受噪声影响程度的特点,在局部方差之间引入模糊区域,通过模糊区域内局部方差的大小来计算相应小波系数的噪声隶属度。根据不同的模糊隶属度在VisuShrink阈值与SUREShrink阈值之间选取去噪阈值,并用软阈值函数消减。实验结果与仿真表明,提出的模糊阈值方法有较好的去噪效果。
In view of the fact that VisuShrink threshold tends to ＂overkill＂ the detail coefficient and the SUREShrink threshold tends to ＂underkill＂ the noise coefficient, a new fuzzy wavelet thresholding for denoising is proposed A fuzzy region in local variance of wavelet coefficient is set based on the characteristic that wavelet coefficient＇ s local variance represents the coefficient dorninanted by original signal or noise. Then the degree of membership by local variance is calculated, the adaptive denoising threshold between VisuShrink threshold and SURFShrink threshold is got, and the coefficient is shrunk by soft threshold. The simulation results show this fuzzy thresholding method is effective to get a better denoising result.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2006年第5期650-653,共4页
Systems Engineering and Electronics
基金
国家自然科学基金(60404022)
河北省教育厅基金(2002209)资助课题
关键词
图像去噪
小波阈值
模糊隶属度
局部方差
image denoising
wavelet threshold
fuzzy membership
local variance
