中文摘要：

根据1／f噪声结构，基于压缩感知（cs）的正交匹配追踪去噪（OMPDN）算法，以小波树结构为分解条件，提取大功率半导体激光器（LDs）中的白噪声及1／f噪声。以小波基作为稀疏基，高斯随机矩阵作为测量矩阵对信号测量并进行CS的重建，滤除白噪声后准确提取1／f噪声信号进行器件参数估计。实验结果表明，本文方法对高斯白噪声混杂的1／f噪声进行估计，比小波变换估计更加逼近真实值。检测估计出的1／f噪声拐点位置及频率指数γ值，为分析LDs可靠性的后续工作奠定了基础。

英文摘要：

According to the structure of 1/f noise,our work is based on compressed sensing and we take the wavelet tree structure as its factor of sparse decomposition to extract white noise and make estimation of 1/f noise parameter. We use the matching pursuit denoising （MPDN） method, set the wavelet base as the sparse base of compressed sensing and use Gaussian matrix as the reconstruction matrix. Our scheme can successfully estimate 1/f noise mixed with white noise and is more precise than the wavelet denoising method. The key parameter γ and the position of inflection point which we estimate can make preparations for the research of LDs reliability.