Coded Aperture Imaging System Based On Compressed Sensing

Computational imaging technology,an advanced science and technology for obtaining target map data.It can improve spatial resolution,spectral resolution,signal-to-noise ratio,etc.It can achieve more efficient data transmission and storage efficiency,meet the needs of multiple fields such as medicine and military.In the compressed sensing theory,only multiple random sub-sampling is needed in the sampling stage,and a small amount of information can be obtained to reconstruct the signal.It can reduce the amount of data collection,reduce the data processing time,and increase the system data processing rate.This work combines coded aperture imaging technology with compressed sensing theory,Digital Micromirror Device is used as the coding template in the coded aperture imaging system,we have designed a coding matrix for compressed sensing,and builds based on compressed sensing theoretical coded aperture imaging system.The main works are summarized as follows:The optical path structure of coded aperture imaging spectroscopy system and coded aperture imaging system is designed and simulated.A novel dual-grating W-shaped spatial optical path structure is designed for the simulation design of the spectral dimension coded aperture imaging system based on the Offner optical principle.In which the lens group is flipped and combined to construct the light splitting system and the light combining system,which saves design time and cost and effectively reduces optical aberrations.At the same time,based on the principle of the spatial dimension coded aperture imaging spectroscopy system and coded aperture imaging system,the simulation design is carried out respectively.The coded aperture imaging technology and the compressed sensing theory are merged with each other.after the coded aperture imaging system with the Digital Micromirror Device as the core is designed,the multiple random sampling matrices are loaded on the device,Both the down-sampled picture collected by the camera and the perceptual reconstruction algorithm can achieve the complete reconstruction of the target.First of all,we compared and analyzed horizontally the orthogonal matching pursuit algorithm,the regularized orthogonal matching pursuit algorithm,the subspace tracking algorithm,and the compressed sampling matching tracking algorithm.Experiments show that the orthogonal matching pursuit algorithm is effective in signal reconstruction,which can be used as the main algorithm for reconstruction.Then,when designing the random coding matrix.In order to ensure the matching with the binarization characteristics of Digital Micromirror Device,the Hadamard matrix consisting of only "-1" or "1" elements is selected as the coding matrix.On this basis,the influence of photoelectric noise,such as Gaussian noise,salt and pepper noise,and speckle noise,on system imaging is analyzed,and the system imaging results when the observation matrix and the coding matrix are not corresponding to the values of the spatial position deviation are analyzed.The data preprocessing and formal processing procedures are designed,which include the construction of the entire coded aperture imaging system from the hardware platform to the PC end data processing.The experimental results show that the target is reconstructed at a low sampling rate after correcting the noise and error of the system.