Autofocusing Raman Probe Based On The Distance Scale Of Monocular Vision

As a non-destructive,non-label rapid material identification method,Raman spectroscopy has been widely used in numerous scientific studies and commercial applications ranging from the characterization of materials to the identification of minerals.The operators of Raman spectrometer have to focus the excitation light on the surface of a sample for the maximum amplitude and the favorable repeatability of the characteristic peaks.Provided that the above-mentioned process is unattended or less direct human involvement,it is known as autofocus.In the light of the existing literature,the main autofocusing methods used in the laboratory can be roughly categorized as the astigmatic method,the centroid method,the sharpness-based method and the intensity-based method.While the above methods have achieved ideal performance on the microscope under certain circumstance,little is known about the way to deal with autofocus in more troublesome situations.This project designed to address the basic issues in autofocus for Raman mapping in vivo is supported by the National Basic Research Program of China(2017YFA0205304)and Shanghai Engineering Research Center for Intelligent Diagnosis and Treatment Instrument(15DZ2252000).We confirm the feasibility of detecting the focus status through the centroid position of the laser spot’s image(CPSI)in theory by the simulation of the Gaussian beam model and the pinhole camera model.Therefore,an autofocusing method based on the distance scale of monocular vision is proposed.The experimental setup contains a handheld Raman spectrograph,two microcontrollers,a camera module and a linear actuator.The experimental results state the influence of the camera extrinsic parameters on the performance is same as the results we simulated by MATLAB.It indicates that the simplified mathematical model can depict the relationship between the CPSI and the focus status well and provides us a detailed guideline of adjusting the performance.After the procedures including image denoising and ambient light compensation,the results of curve calibration show that the working distance of our method is at least 43.5 mm and the detection threshold in the linear range is around 1 μm.Both the experiment and the simulation demonstrate the deviation between the measured value and the true value caused by the non-normal incidence is inevitable but the deviation tends to be minimal at the focus.Lastly,an autofocus-free handheld Raman spectrograph utilizes this method to autofocus the 95% alcohol in the centrifuge tube successfully and the spectral reproducibility is improved.Our results may pave the way to a novel autofocusing approach for Raman mapping in vivo.