Binary Pure Phase Encoding Compressive Imaging in Frequency Domain

Super resolution (SR) is considered as one of the “holy grails” of optical imaging and image processing. The introduction of compressive sensing theory presents a novel super-resolution reconstruction method from a single low-resolution image, which can avoid the requirements for the multiple sub-pixel images of traditional superresolution method. Analyzing the requirements of the similarities and differences between compressed sensing measurement matrices and optical imaging systems, a binary phase encoding compressive imaging method based on the 4-f optical architecture is presented, with the phase in the frequency domain randomly modulated, which can achieve super-resolution reconstruction from single low-resolution measurement images obtained with single exposure conditions, no other additional information collected. Binary phase mask is much easier to implement than random phase mask with uniform distribution, which is a more viable scheme for physical realization of compressive imaging. Simulation experiments demonstrate that the proposed method can effectively capture compressive measurements and implement super-resolution reconstruction in a single shot condition. Furthermore, another experiments show that this method is also more applicable to large-scale image reconstruction compared with random demodulation (RD) proposed by Romberg in the reconstruction time, and more practical in the sampling scheme than RecPC method proposed by Yin.