Sparse Modeling for Spectrometer Based on Band Measurement

In typical spectrometric measurement systems, a high-resolution spectrum is obtained directly via sequential observations with a narrow slit-like measurement window at the expense of sensitivity. In this paper, we propose a novel spectrometric method applicable to these typical spectrometric systems: a multiplexed low-resolution measurement with a wide measurement window, band measurement (BM), is combined with sparse-modeling-based post-processing to obtain the original high-resolution spectrum. BM is expected to improve the measurement signal-to-noise ratio because of the increase in the sample quantities reaching the detector by widening the measurement window. BM has the significant practical advantage that it can be easily implemented in spectrometric measurement systems without device alterations. To evaluate the effectiveness of our proposal both theoretically and experimentally, we formulate the sparse-modeling-based post-processing in the proposal in terms of the least absolute shrinkage and selection operator (lasso) and perform a theoretical analysis and simulation studies concerning the resulting spectrometric method named BM-lasso. In the theoretical analysis, we derive density evolution equations for belief propagation on the BM-lasso model and obtain the expected errors of estimators of BM-lasso. The numerical evaluations of the theoretical analysis result revealed that BM-lasso achieved lower mean square error than the conventional measurement method under the...
Source: IEEE Transactions on Signal Processing - Category: Biomedical Engineering Source Type: research