Exploring exoplanets using stellar coronagraph requires coronagraph with a contrast of 10<sup>-10</sup> or even lower, because the difference in light intensity between the planet and its parent star is very large. To this end, we optimized the coronagraph imaging system with Four-Quadrant Phase-Mask (FQPM) proposed by D. ROUAN et al in 2000. The FQPM has the advantages of high extinction efficiency for the coherent light from the main source, low sensitivity to atmospheric turbulence and large dynamic range. This paper proposed an apodizer with continuous transmission and a Lyot stop optimized in conjunction with the apodizer for FQPM coronagraph, which enhance the nulling ability of FQPM to achieve a high contrast ratio of 10<sup>-12</sup> at 1.25λ/D and a contrast of 10<sup>-13</sup> at larger distance. Moreover, this optimization method can optimize the non-circular symmetric mask, that is, the optimization method is two-dimensional, rather than one-dimensional optimization in the case of circular symmetry. Then we compare the joint optimization with the optimization of the apodizer only, the results show that the former has better diffraction suppression and concentrating ability than the latter, which makes the energy more concentrated and the peak signal obtained on the detector is stronger. In the follow-up work, we will continue to complete the FQPM coronagraph system, such as adding adaptive optical to correct wavefront distortion, adding considerations for manufacturing precision of optical components and so on.
Different degradation factors such as Poisson noise, blurring effect, different contrast and different reflectivity and so on will impose severe influences on the imaging process of the non-cooperative space targets with low light intensity and the corresponding image quality is usually poor. In this paper, a two-step reconstruction framework based on compressed sensing (CS) theory is proposed to deal with these degradation factors to improve the quality of the space target images. The proposed algorithm is divided into two steps, the first step is standard compressed sensing based reconstruction, and the second step is super-resolution based on the theory of compressed sensing. Specifically speaking, when the sparsely sampling are obtained, the total variation augmented Lagrangian alternating direction algorithm (TVAL3) is first used to recover the 2D image, which only obtain 25% of the number of pixels in the original image instead of all the pixels in the traditional sampling. Subsequently, the single-frame image super-resolution reconstruction is performed on the captured 2D image, and the super-resolution algorithm based on the dictionary learning is used to realize super-resolution reconstruction, which makes the image resolution doubled.