基于小波域DCT变换的杂波抑制方法  

Clutter suppression method based on DCT in wavelet domain

作　　者：周宁[1] 李晓峰[1] 李在铭[1] 

机构地区：[1]电子科技大学通信与信息工程学院,成都610054

出　　处：《计算机应用研究》2010年第1期374-376,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60873263);国家"863"计划资助项目(2009A01Z215)

摘　　要：针对复杂背景中微弱运动目标检测困难的问题,提出了一种基于小波域DCT变换的背景杂波抑制方法。该方法根据背景杂波和运动目标的不同频率特性,采用低频小波子带频域滤波的方法得到有效抑制背景杂波的残差图像,从而达到抑制背景杂波的目的。该方法首先对原始图像进行小波变换,接着对低频小波子带进行二维DCT变换,再用高斯低通滤波器对DCT变换结果进行滤波,然后对滤波结果进行IDCT变换,最后对滤波前后的低频小波子带作差分处理,对差分结果进行小波逆变换。实验结果表明,该方法处理后得到的残差图像呈现出很好的高斯性和独立性,并且目标邻域信杂比(SCNR)的平均增益比图像直接频域滤波的目标邻域信杂比平均增益提高2 dB以上,算法性能明显优于传统的图像频域滤波算法。Aiming to detect dim moving targets in complex background, the low wavelet belt was frequency filtered to suppress background clutter and get the residual image according to the different frequency characteristics of background clutter and moving targets. Firstly,performed the wavelet transform,and operated DCT of two dimension to the low wavelet belt （LL） , filtered the lower frequency components of LL by Gaussian lower pass filter, and then did the IDCT to the filtering results. Operated the difference process between the preand after filtering LL. Finally,carried out the inverse wavelet transform. The experiment results show that the residual image obtained by this method has very good Gaussian normality and independence, and the average gain of the target＇ s neighbor SCNR （signal-to-clutter-noise ratio） is improved above 2dB, compared with the image frequency filtering algorithms. So the method has better performance than conventional image frequency filtering method.

关 键 词：背景杂波抑制 微弱运动目标 小波域离散余弦变换 目标邻域信杂比 

分 类 号：TN911.73[电子电信—通信与信息系统]