Research On Coded Aperture Spectral Imaging Technology Based On Compressed Sensing

Spectral imaging combines imaging technology with spectroscopy.It can detect spectral information while obtaining spatial information of the target scene.Therefore,the spectral imaging technology has a wide range of applications in various fields.With the development of science and technology,the requirements for spectral imaging technology in various fields are becoming higher and higher.High space,spectral resolution,high optical utilization rate,high signal-to-noise ratio,and high efficiency storage and transmission requirements bring new challenges and it also point the new directions for spectral imaging technology.In the current application of spectral imaging technology,the main methods are using whisk broom scanning,push broom scanning or interferometric methods to obtain target information.However,due to their own principle limitations,the improvement of one performance index will inevitably lead to the decline of another index.The traditional spectral imaging technology based on the sampling method of Shannon sampling theory has high requirements on the sampling frequency.In order to obtain higher imaging quality,the sampling frequency is bound to increase,which results in the increasing demands for system complication and technical requirements.In addition,the increased resolution has higher requirements for photodetector components.Therefore,the researches on new spectral imaging mechanisms and new sampling theory are of great significance for the further application of spectral imaging technology.In recent years,the computational imaging,as a branch of modern optics,has attracted more and more attention due to its advantages of opto-mechatronics integration design.Computational spectral imaging technology,as a category of computational imaging,uses compressed sensing sampling theory which is different from Nyquist sampling theory,to recover complete spatial information and spectral information with fewer measurement data.It has become one of the important developing directions of spectral imaging technology.Among them,the most promising technique,coding aperture spectral imaging technique,has attracted the attention of researchers in this field.Coded aperture spectral imaging technique has the advantages of snapshot imaging,high luminous flux,high signal-to-noise ratio and low sampling frequency.The coded aperture,as one of the key components of the system,requires more optimization and improvement before its engineering application.The resolution of the coding aperture is firstly analyzed.Then the coding aperture is optimized and a novel algorithm which can directly perform spectral unmixing utilizing the compression measurement results is put forward.Finally,a new coded aperture mode is constructed.In this dissertation,the main research contents are as following:Firstly,the spectral imaging technology and the basic theory of spectral imaging are deeply studied.Through studing and discussing computational spectral imaging techniques,the theory for compressed sensing sampling has been systematically completed.Furtherly,by deriving the mathematical model and analyzing the key components,the study of coding aperture spectral imaging technique is completed.Secondly,the problems encountered in the practical application of the coded aperture spectral imager are analyzed in detail.By establishing the mathematical model of the coded aperture snapshot spectral imaging system(CASSI),the factors such as the influences of the mismatch between the coding template and the detector resolution are analyzed.Aiming at the fact that the resolution of the coding template is higher than the resolution of the detector,it is proposed to introduce the super-resolution technology into the CASSI system to achieve super-resolution spectral imaging based on compressed sensing.For the case where the resolution of the coding template is lower than the resolution of the detector,a coding aperture with gray level grading is proposed to achieve high resolution coding mode,which can ensure the resolution of the coded aperture spectral imager.The effectiveness and feasibility of the proposed methods are verified by experimental tests.Thirdly,the spectral unmixing of the coded aperture snapshot spectral imager is complex and slow.With the sparse representation and the sparse estimation of end-battery abundance,an algorithm for directly performing spectral unmixing on the obtained measurement results is proposed,which is utilized to replace the traditional complicated process of reconstructing the data cube and then performing spectral unmixing.Experiments are carried out using the designed coded aperture.The results demonstrate that the optimized coded aperture and the new unmixing algorithm have faster spectral unmixing speed and higher precision.Lastly,the structure of the coded aperture snapshot spectral imager is optimized by using a color optical filter array instead of the coded aperture and combining it with the detector.Based on the above method,a snapshot color compression spectrum imaging system is proposed.The mathematical modeling and reconstruction algorithm design of the system are carried out.Computer simulation results illustrate that the optimized system has higher reconstruction accuracy than traditional coded aperture spectral imaging systems.