基于多小波变换及综合阈值的图像去噪方法  

Image de-noising method based on multi-wavelet transform and synthesis threshold

作　　者：陈苏婷[1] 吴钦章[1] 

机构地区：[1]中国科学院光电技术研究所

出　　处：《红外与激光工程》2007年第1期139-142,共4页Infrared and Laser Engineering

基　　金：国家863高技术课题资助项目(2001AA131010)

摘　　要：提出了一种基于多小波变换的综合阈值图像去噪方法。该方法通过对含噪图像进行多小波变换,克服了单小波变换中无法同时满足正交性和对称性的缺点。同时,在两种经典的硬、软阈值处理方法基础上,提出了一种综合阈值处理方法。将该阈值方法和多小波变换相结合,根据多小波分解后的能量分布特性,在不同尺度的子带选择不同的最佳阈值,有效地提高了重构图像质量。并且对阈值进行向零收缩处理,防止有用信息当作噪声滤除。实验结果表明,相对于传统的硬、软两种阈值处理方法,文中的去噪方法在输出信噪比和主观视觉效果上都有明显改善。An image de-noising algorithm based on multi-wavelet transform and synthesis threshold method are presented. By adopting multi-wavelet transformation, the algorithm overcomes the limitation of single wavelets that can＇t have orthogonality and symmetry simultaneously. Meanwhile, a synthesis thresholding method is proposed based on two classical thresholding methods, hard-thresholding and softthresholding. According to wavelet coefficients energy distribution in different subbands, different thresholds are obtained in different subbands to improve reconstructed image quality significantly. The algorithm is also processed toward zero shrink to avoid useful signal being filtered out as noise. The experimental results show that, compared with two classical methods, the algorithm combining multi-wavelet transform and new threshold method greatly improves the SNR value and subjective visual effects.

关 键 词：图像去噪 多小波变换 综合阈值 

分 类 号：TN911.73[电子电信—通信与信息系统]