摘要
为研究激光主动成像中散斑噪声的抑制问题,提出一种基于信号子空间TDC(Timedomain constrained)的散斑去噪方法,并搭建一套基于距离选通ICCD的激光主动照明系统进行实验验证。首先使用同态变换将乘性噪声变为加性噪声,然后利用小波变换估计噪声的协方差;接着对含噪图像进行奇异值分解并估计信号子空间的维数,根据该维数对无噪图像的协方差矩阵进行特征值分解,计算出滤波估计矩阵。将滤波估计矩阵与含噪图像卷积,最后做同态逆变换,得到降噪后的图像。结果证明本文的去噪方法拥有比经典的Lee、Frost和Kuan算法更好的散斑噪声抑制效果,同时计算时间明显缩短。
To investigate the suppression of the speckle noise in laser active imaging, a denoising method based on Time-Domain Constrained （TDC） in signal subspace is proposed and a laser active imaging system based on range-gating ICCD is constructed for experiment. First, homomorphic transformation is performed to convert the multiplicative noise to additive noise. Second, wavelet transformation is performed to estimate the covariance of speckle noise. Third, the noise image is decomposed into signal subspace and noisy subspace, and singular value decomposing is used to estimate the dimension of the signal subspace. The covariance of clean image is decomposed using eigenvalue decomposing, and denoising estimating matrix is computed. Fourth, the denoising estimating matrix is convolved with noisy image. Finally, the inverse homomorphic transform is carried out to get the denoised image. Experiment results indicate that, compared with classical Lee, Frost and Kuan filtering, the proposed method has advanced denoising performance and consts less computation time.
出处
《吉林大学学报（工学版）》
EI
CAS
CSCD
北大核心
2015年第4期1347-1352,共6页
Journal of Jilin University:Engineering and Technology Edition
基金
吉林省科技发展计划项目(20126015)
