Research On Ultrafast Imaging Technology Based On Compressed Sensing

Ultrafast imaging technologies enable picosecond or higher temporal resolution observation of ultrafast phenomena,which is of vital importance to a number of research areas covering physics,chemistry,and life sciences.Compressed sensing samples high-dimension information into lower dimensions,showing the potential to boost the performance of ultrafast imaging where high-dimension information is inherent.On the one hand,single-photon-avalanche-diode(SPAD)arrays are able to achieve single-photon sensitivity and picosecond-level temporal resolution,it has broad applications in ultrafast imaging where weak repeatable signals are present.However,existing SPAD arrays exhibit low pixel count,limited temporal resolution(due to on-chip TDC),low fill factor,and "screamer pixels",precluding its application in high temporal resolution and large field-of-view scenarios.On the other hand,100fs-order active framing imaging can be achieved based on the time-wavelength relation of linear chirped pulses.However,spatial multiplexing can not be employed with this method,yielding a low frame number or low photon efficiency.This work exploits compressed sensing to overcome the above-mentioned disadvantages of the two methods.The main contents are the following:1.Time-resolved single-photon imaging based on compressed sensing.The combination of single-photon imaging and TCSPC technique to achieve time-resolved imaging using a single SPAD is proposed.It is observed in experiment that the laser pulse propagated with 8ps time intervals.An optimized coding method is used to improve the efficiency of the imaging system,where the target scene can be reconstructed with a sampling rate of 0.5.It is also shown that the key information in the target scene can be reconstructed with a sampling rate as low as 0.05.This technique avoids the above-mentioned problems with existing SPAD arrays.Besides,it yields a better temporal resolution and a tunable spatial resolution.2.Ultrafast framing imaging based on compressed sensing.This work develops a theoretical model for compressed ultrafast framing imaging.An algorithm based on dispersion-corrected 3D-total-variation generalized alternating projection is proposed,which yields more robust performance.A compressed framing imaging system is built experimentally,with which 100fs-order temporal resolution imaging with 30 frames has been achieved.Theoretical analysis shows that the proposed system yields tunable temporal resolution from 100 fs to ps with tens of frames and mega-order number of pixels.3.Photorefractive imaging based on compressed ultrafast framing.Ultrafast response chips made of low-temperature grown AlGaAs(LT-AlGaAs)are utilized to realize imaging of picosecond level temporal resolution and 40lp/mm @MTF=0.1spatial resolution.We analyzed the spatial performance of ultrafast response chips via X-ray.Experiments show that LT-AlGaAs enable high spatial resolution(≥35lp/mm@MTF=0.1),large field-of-view(6.7mm×6.7mm)imaging with 120:1 dynamic range of incident X-ray.This work proposes an experimental design that employs a reference grating and a coated grating on the ultrafast chip as the basis grating for photorefractive imaging based on compressed ultrafast framing.This method enables ultrafast passive imaging.This work utilizes compressed sensing theory to improve the performance of a couple of ultrafast imaging modalities.Two new methods are proposed,namely the time-resolved single-photon imaging based on compressed sensing and ultrafast framing imaging based on compressed sensing.Experiments have been conducted which show the improvement in the imaging performances.The theory and experimental designs have also been discussed for photorefractive imaging based on compressed ultrafast framing.Ultrafast imaging using LT-AlGaAs chip and its X-ray dynamic range test are performed as parital proof of concept.