Research On Spectral Encoded And Signal Weighted Fourier Ghost Imaging

The optical image is one of the main means for human beings to recognize the world.Compared with other cognitive approaches,the optical image is more intuitive,comprehensive,and rich.In recent years,a new imaging technology,named ghost imaging or correlated imaging,which is quite different from the traditional imaging technology via array detectors,has attracted the continuous attention of scholars at home and abroad.Compared with the conventional imaging technology via array detectors,the imaging device of the ghost imaging is the single-pixel detector(such as photomultiplier tube,et al),and combines the optical modulation device and the corresponding recovery algorithm to obtain the image of the target object(s).Sometimes,ghost imaging technology is often called single-pixel imaging technology.The spectrum selectable range of single-pixel detectors is broader than that of array detectors,which makes the ghost imaging technology have potential imaging advantages in non-visible light or the spectrum band that array detectors can not image.In addition,combined with high-speed photoelectric detection systems and short-pulse laser source,there are some significant advantages in long-distance and high longitudinal resolution of three-dimensional imaging for ghost imaging.At present,ghost imaging technology has illustrated unique application prospects in the fields of spectral imaging,terahertz imaging,three-dimensional imaging,gas leakage imaging,target tracking,and so on.Unfortunately,until now,there is still a certain gap between ghost imaging and traditional array imaging in the field of some imaging parameters of spatial resolution,quality,efficiency,etc.How to improve these performance parameters of ghost imaging technology,lots of researchers had made great efforts,and fruitful results have been achieved.Here,based on the previous studies,this thesis mainly focuses on the research of multispectral ghost imaging and Fourier ghost imaging technology.Aiming at the shortcomings of existing technologies,some novel strategies are proposed to overcome the existing issues.The corresponding theoretical analysis,numerical simulation and experimental verification are carried out in this thesis.There are six chapters in this thesis.The first chapter mainly summarizes the development and application of ghost imaging technology;the second chapter introduces the image restoration methods of ghost imaging technology;the last chapter summarizes and prospects the research topics;and the third,fourth and fifth chapters are the main work of this thesis,which are summarized as follows:(1)Multispectral ghost imaging generally makes use of spectral splitting approach to obtain the spectral component images of the imaged object by using multiple single-pixel detectors simultaneously or using a single-pixel detector in time-sharing,and then the corresponding multispectral image is fused by the spectral component images.These technologies suffer from complexity,a large amount of data,and low efficiency.Aiming at the shortcomings of the traditional multispectral ghost imaging technology,the spectral encoded computational ghost imaging technology of the orthogonal modulation illumination strategy is proposed.The orthogonal multispectral modulation illumination patterns are generated through fused the deterministic orthogonal Hadamard basis matrices with the spectral encoded matrices.A projection system based on digital light projection technology is employed to illuminate the imaged object,and a single-pixel detector is used to collect the back-scattered light of the imaged object,which is then discretized and stored.The process of the multispectral image restoration process is as follows:firstly,the mixed spectral image of the imaged object is recovered from the detection signal by the evolutionary compressed iteration technology;secondly,by means of the orthogonality of the spectral encoded matrices,the sub-sampled spectral component images are obtained by the dot product operation of the reconstructed mixed spectral image and the spectral encoded matrices,respectively,and then the group sparse compressed sensing algorithm is employed to reconstruct the fully-sampled spectral component images.The corresponding multispectral image is fused by the fully-sampled spectral component images.The spectral encoded computational ghost imaging technology of the orthogonal modulation illumination is analyzed theoretically,and the numerical simulation and experimental verification are carried out.Besides,the comparative experiment with the traditional multispectral ghost imaging technology is also carried out.The simulation and experimental results show that the proposed method is effective.The constructed orthogonal modulation illumination patterns effectively overcome the redundancy of the random modulation patterns,which improve the efficiency of multispectral ghost imaging.The proposed method simplifies the system configuration of multispectral ghost imaging,reduces the amount of data collected and the recovery time.(2)The prior knowledge is that in the Fourier transform domain,most of the energy of natural scenes is concentrated in the low-frequency region.According to this concept,the Fourier patterns in the low-frequency region can be employed in ghost imaging to modulate the illumination light source or the reflected or transmission signal of the imaged object.This strategy can reduce the number of samples and illumination patterns,and thus improve ghost imaging efficiency.However,the currently popularly used spatial light modulator,such as the digital micromirror device has a low refresh rate to the grayscale Fourier patterns,which restricts the further development and application of traditional Fourier ghost imaging technology.To address this issue,the signal weighted Fourier ghost imaging technology is proposed.The grayscale Fourier basis modulation patterns are converted to the corresponding binary modulation patterns sequence.The weight of the detection signal corresponding to the binary modulation patterns replaces the weight of the illumination time of the corresponding binary modulation patterns.This strategy will significantly improve the efficiency of Fourier ghost imaging based on a digital micromirror device.Technically,the grayscale Fourier basis patterns are decomposed into the corresponding binary modulation patterns sequence to modulate the light source according to the principle of decimal to binary.This process takes full advantage of the DMD's high refresh rate for binary patterns.The Fourier spectrum(of the imaged object)of the corresponding grayscale Fourier basis patterns are equivalent to the weighted sum of the detected signals to the corresponding binary patterns sequence.And then an inverse Fourier transform is applying to the captured Fourier coefficients to recover the image of the imaged object.The method of the proposed signal weight Fourier ghost imaging is analyzed in theory.And the image restoration quality of different target objects under different quantization levels of the Fourier basis patterns which are changed into the corresponding binary modulation patterns sequence are evaluated by numerical simulation.The experimental verification work is carried out based on theoretical and numerical simulation.The experiments are carried out on static and dynamic objects,respectively.And video imaging with a frame rate of about 9 frames is achieved.The proposed method promotes the development of Fourier ghost imaging technology.(3)Time of flight three-dimensional ghost imaging LIDAR(Light Detection and Ranging)has the ability to simultaneously obtain the distance information and the image of the imaged object in the imaging scenes.It has a broad application prospect in remote sensing,target detection recognition,and other fields.Here,the signal weighted two-dimensional Fourier ghost imaging technology is extended to three-dimensional Fourier ghost imaging.Combination with the time of flight method,time of flight three-dimensional Fourier ghost imaging is proposed.The imaging process and image restoration method of time of flight three-dimensional Fourier ghost imaging are analyzed.The corresponding experimental study of the time of flight three-dimensional Fourier ghost imaging is carried out.The Fourier spectrum slices of the imaged scenes are calculated by the weight of the detection signal corresponding to the binary modulation patterns,then,the slice images,intensity images and distance maps of the imaged scenes are reconstructed or calculated respectively by applying an inverse Fourier transform to the different spectral components of the captured Fourier spectrum slices.The experimental results show that the top 25%of the frequency spectrum components can achieve clear imaging of the intensity images and distance maps.In an experiment of the target objects obscured by the netting obstruction,the results show that the quality of the intensity images and distance maps restored by the top 25%frequency spectrum components is significantly better than the quality of the intensity images and distance maps retrieved by the whole frequency spectrum components.The preliminary experimental results reflect that time of flight three-dimensional Fourier ghost imaging has potential technical advantages in removing netting obstructions.