Single Pixel Camera Real-time Reconstruction Imaging Method Based On Coding Light

Compressed sensing(CS)theory is a sampling theory proposed in recent years,a new theory that was put forth in the field of signal processing and information theory.Due to it breaks the traditional Nyquist sampling theorem,CS has been widespread concern academia and industry.And it has been successfully applied to wireless communications,medical imaging,earth science and many other fields.Single-pixel imaging technology is to prove the CS theory when it was first proposed,because its theoretical background is CS theory and it can be achieved with less sample volume while the accuracy of image recovery is the same,in addition it can be easily extended to other wavelengths,therefore it has broad application prospects in both civil and military.However,the reality that a single pixel platform built mostly at the laboratory stage is still a great distance away from industrialization.The reason is that the platform is too complex to build.Moreover,the image reconstruction accuracy and speed compared to traditional imaging modalities is still insufficient.To solve these problems,this thesis sets up a single pixel imaging platform,and puts forward relevant measurement matrix optimization algorithms and optimization algorithms to accelerate recovery,to achieve the existing single-pixel imaging technology overall improvement.This paper includes the following aspects:1.In recent years,by researching and comparing the characteristics of a single pixel platform development,we overcome many shortcomings of traditional single-pixel imaging device,and build a flexible single-pixel imaging platform with the portable projectors and light intensity detector which achieved without Single-pixel imaging prism and aperture.Comparing to other single-pixel imaging platform,ours has a broader applicability.2.Measurement matrix which is important in C S coding process is optimized by hierarchical model proposed by our thesis.Then we construct different resolution measurement matrix and apply the prior acquired from low-level image guide information to the high-resolution image reconstruction process,which greatly improve the accuracy and speed of image reconstruction.3.In order to achieve a real-time image reconstruction,this paper proposes a recovery accelerated optimization algorithm,which transforms the most time-consuming iterative process into a simple sum of multiplying.The proposed L1-min algorithm greatly reduces the recovery time.4.Through experiments,we compare proposed measurement matrix optimization algorithm and L1-min acceleration optimization algorithm with existing mainstream algorithms and analysis advantages and disadvantages of methods by using two indicators including reconstruction error and time consuming.