论文部分内容阅读
针对由光谱仪器测得的拉曼光谱数据经常会受到随机噪声和仪器误差等的影响而导致低分辨率的问题,文中提出了一种能在恢复光谱结构的同时又能抑制光谱噪声的方法,即基于LaplacianMarkov约束的数据加权光谱反卷积模型。该模型将退化光谱中恢复真实光谱的问题转化为最大后验概率的求解问题,推导出了拉曼光谱恢复的变分模型。模型中利用Laplacian-Markov作为光谱数据的光滑性先验,提出加权光谱反卷积来恢复退化的光谱,并使用分裂Bregman迭代法求解。文中对该算法利用实验数据进行了验证,实验表明该方法既能恢复退化光谱细节又能抑制光谱噪声,并且求解速度快,有较强的实用价值。
Raman spectroscopy data measured by the spectrometer is often subject to low resolution due to random noise and instrumental errors. In this paper, a method that can restrain the spectral noise while restoring the spectral structure is proposed. The data-weighted spectral deconvolution model based on LaplacianMarkov constraint. The model transforms the problem of recovering the real spectrum in the degradation spectrum into the problem of solving the maximum a posteriori probability, and deduces the variational model of Raman spectrum restoration. Using Laplacian-Markov as a smooth priori for spectral data, we propose a weighted-spectral deconvolution to recover the degraded spectra and solve them using the split-Bregman iterative method. In this paper, the algorithm is verified by experimental data. Experiments show that the proposed method can not only restore the details of degraded spectra but also restrain the spectral noise, and the solution speed is fast and has strong practical value.