Multispectral Imaging Based On Digital Micro-mirror Device

Spectral imaging combines imaging and spectroscopy.It contains the spatial and spectral information and has wide applications.This thesis mainly studies the Digital Micro-mirror Device(DMD)-based spectral imaging.The contents and major results are shown below.1.The spectral imaging technology and its classifications are introduced.The coding spectral imaging is our study focus.From the research status of DMD-based spectral imaging,two coding mode of DMD is summarized,including spectral coding and spatial coding.2.The basic theories of spectral coding and spatial coding are described respectively.And the diffraction effect of DMD is analyzed.According to the coding feature and affecting factors,a DMD-based spectral imaging with dual-grating structure is designed.3.Hadamard transform spectral imaging is implemented based on the designed system.To simplify the traditional calibration process,a calibration method using image matching is proposed.This method obtains the physical location of coding pattern and does not need recalibration for changed patterns.Besides,it calculates the gray coding values for inter-coding pixels.The reconstructed intensity is more accurate using gray value than using 0/1 coding vaule.4.A new Fourier spectral imaging method is studied.The DMD sinusoidal patterns are used for Fourier transform.This method has several advantages.First,comparing with Hadamard transform spectral imaging,it reduces sampling time for compressed measurement.Second,it has flexible spectral resolution by using different number of DMD patterns.Third,the system has no moving parts and the reconstructed spectrum is in wavelength domain directly.5.The compressed sensing snapshot spectral imaging is analyzed.Its snapshot can be used for moving object,but brings poor spatial quality.To improve the spatial quality,a micro-lens array with different filters is introduced.CCD detects a group of images with same spatial information but different wavebands.The reconstruction for sub-wavebands will improve the spatial quality and it is validated from simulation.