Research And Application Of Spectral Bandpass Correction Technology For Micro Spectrometer

Spectrometers are indispensable equipment for spectra analysis.However,traditional spectrometers are not only bulky and expensive,but also very complicated to use.It is difficult to meet the needs of researchers in the field.Therefore,the research focus has shifted to the miniaturization and intelligence of spectrometers.Although the existing micro spectrometers on the market have made great progress in volume,price and speed,the measurement performances are not satisfactory,and in serious cases,the entire distribution of the spectra will be changed.This has a very big impact on studies that need to be completed by analyzing spectral data.The reasons for the above problems are: the wavelength calibration accuracy is not high enough,the input photons in the photoelectric conversion process are not equal to the output electrons,the measurement process contains noise,the non-target beam reaches the image surface,and the light is scattered as it enters the micro spectrometer system through the slit.Among these,the largest influence is the spectral broadening caused by scattering.In order to reduce the analysis error and improve the accuracy and precision of the measurement results,we need to perform bandpass correction on the initial spectral data obtained by the measurement.This paper first introduces the cause of spectral broadening and its impact on the spectral analysis.And discusses five classic bandpass correction methods in detail,including: Stearns and Stearns(S-S)method?Ohno method?Differential Operator(DO)method?Richardson-Lucy(R-L)method and Levenberg-Marquardt(L-M)method.At the same time,the advantages and disadvantages of each correction method are summarized,and the methods for evaluating the spectral correction effect are given.Bandpass correction is actually a reverse solution process.Here we propose a correction method based on Maximum a Posteriori Estimation Method(MAP),thus the bandpass correction problem is converted into a probability prediction problem.Then organized three sets of simulation experiments(no noise,with photoelectron noise,and with random noise),and compared the correction effects of these methods.Both the effectiveness and applicability of the Maximum a Posterior Estimation Method(MAP)in spectral bandpass correction are verified.Finally,it is applied to the wavelength position correction and the color measurement of LED based on micro spectrometer.In order to improve the credibility of the measurement results,the wavelength calibration must be completed before the micro spectrometer is officially put into use.The traditional calibration is done by manual peak finding and polynomial fitting higher than fifth order.The calibration accuracy is difficult to meet the experiment claim.In this paper,the Maximum a Posterior Estimation Method(MAP)and Voigt linear spectra model are used to achieve wavelength calibration.The experimental results show that the calibration error of the traditional method is about 0.4nm,and the calibration error of our method is about 0.1nm,which means that the calibration accuracy has been improved.Based on the micro spectrometer after calibration,a set of LED color measurement system is built.The light source to be tested is the LED lamp produced by CREE company,including three white LEDs at different CCTs and three monochromatic lights.The direct error between the chromaticity coordinates obtained after correction and the reference coordinates is about 0.0002.Compared with the unprocessed chromaticity coordinates,the error is reduced by an order of magnitude,which further verifies the effectiveness of the proposed method.