Image clarity of membrane diffractive imaging system using total-scale Retinex

Transmissive membrane diffractive optic can be used in space optical telescope or long-distance sensing to reduce the size and mass of imaging system. However, images of membrane diffractive imaging system are inevitably affected by diffraction efficiency and atmospheric propagation, which leads to low utilization ratio of light energy and image blur. Clarity of this degraded data is a challenging task. In this paper, an effective post-processing method based on total-scale Retinex (TSR) is proposed to address this problem. First, we design a self-adaptive scale surround function model that is guided by the optical transmission rate to estimate illumination component. Then, we embed the estimated-illumination into Retinex model for directly estimating the final clarity component. Moreover, the proposed method is capable of quasi real-time processing video using temporal coherence. Our method extends the multi-scale Retinex model while it avoids color artifacts in transmission model that can appear due to incorrect depth estimation. The experimental results show that our approach obtains better performance than the other state-of-art methods with respect to contrast enhancement and color-cast correction.