Innovative Remote Sensing Imaging Method Based on Compressed Sensing
Li Sheng-liang,Liu Kun,Zhang Feng,Zhang Li,Xiao Long-long,Han Da-peng
DOI: https://doi.org/10.1016/j.optlastec.2014.03.019
2014-01-01
Abstract:Compressed sensing (CS) has good application prospects in remote sensing imagery. In particular, CS theory can be used to alleviate the burden for remote sensing data transmission and recover scenes at high resolution. However, the application of CS theory to practical remote sensing imaging systems involves some key challenges: (i) Many random projections cannot be implemented in practical systems. (ii) These random projections cannot change after the linear optical system is fixed. (iii) Some traditional imaging systems, such as single-pixel cameras, are unfit for spacecraft. Therefore, innovative imaging systems must be designed for remote sensing imaging. In this paper, we review CS theory, and present a remote sensing (RS) imagery system based on CS. The video sequence measurement model is introduced. Three CS optical architectures, namely, single-pixel, coded aperture, and photon-superimposition, are presented. We then design a new imaging architecture that uses a phase modulator and subframe superimposition, and introduce phase modulator matrix in correspondence. A coupled reconstruction model for recovering video is proposed. Through numerical simulations, we demonstrate the effectiveness of the optical architectures and compare it with two traditional architectures (coded aperture, and photon-superimposition). The effectiveness of coupled models is compared with traditional single models. We can conclude that, the proposed imaging architecture can precisely assess resolution and increase field of view. Furthermore, the presented coupled model can easily improve accuracy and speed of video reconstruction.