Single Pixel Camera Imaging Svstem Based On Compressed Sensing Theory

In the process of conventional image transmission, there is only a little effective data to be saved after compressing the collected data. Undoubtedly, that causes a great waste of collecting resources, storage resources and the cost. In recent. Candes, Romberg, Tao and Donoho proposed the theory of compressed sensing, that provides the theoretical basis to solve the front waste. The single-pixel camera imaging system based on compressed sensing theory only needs one photon detector, and collecting and compressing image can be carried out simultaneously, which greatly reduces the size,cost and complexity of the image system.This paper further researches the structure of the single-pixel camera system and the system is divided into three parts, which are optical path modulation module, data acquisition module and image reconstruction module.1) Optical path modulation moduleThe state of the micromirror on DMD chip is set according to the measurement matrix, so that the optical image on focus plane can be collected linearly. We mainly research the sparse binary random matrix, binary permuted block diagonal matrix and partial discrete walsh matrix. Using the reconstruction algorithms, we did the simulated and real imaging experiments.2) Data acquisition moduleIn this module,photomultiplier tube is used to measure the total energy of the collected light and data acquisition card is used for A/D conversion, and then using programmer to save the data.3) Imaging reconstruction moduleAccording to the measurement matrix and collected measurement vector, the object image can be reconstructed by the reconstruction algorithms accurately. We mainly research gradient projection for sparse reconstruction algorithm and TVAL3algorithm. Based on these two algorithms, we develop the image reconstruction software by vs2010.Different measurement matrix and reconstruction algorithms had been analyzed and compared in this paper. The condition for the measurement matrix composed only0and1to reconstruct the object image accurately had been proposed. The image reconstruction software provides the software foundation for the practical application of the single-pixel camera. The practical experimental platform had been built, and two groups of experiments had been done. The first group used sparse binary matrix and gradient projection for sparse reconstruction algorithm and the second group used partial discrete walsh matrix and TVAL3algorithm. The experimental results show that the single-pixel camera can achieve high resolution image and has the anti-interference ability. And the reconstruction speed of the second group is much faster than the first one and the reconstruction effects of the second group are much better than the first one. In second group, when the measurement rate is5%, it can see the contour of the object image from the reconstructed image. When the measurement rate is10%-20%, it can reconstruct the object image accurately. So for single-pixel camera, the partial discrete walsh matrix can be used as the measurement matrix and TVAL3algorithm can be used as the reconstruction algorithm, when it is used in practices.