Photon Counting Integral Imaging Using Compound Photon Counting Model and Adaptive Parametric Maximum Likelihood Estimator

In this paper, a statistical approach based on an adaptive parametric estimator is proposed for the three-dimensional (3-D) reconstruction of objects under photon-starved conditions. In photon counting integral imaging system, 3-D objects having small number of photons can be visualized by the prior-based statistical estimation. However, improper prior constrains can lead to inaccurate reconstruction results. The adaptive parametric Maximum likelihood estimator (MLE) using a compound photon counting model is proposed to visualize the photon-limited 3-D objects. Through maximizing a likelihood function with pixel-based adaptive information, the number of photons for reconstructed pixels is estimated. Variance stabilizing transformation combined with Block-matching and 3-D filtering algorithm is also applied to enhance the photon counting elemental images captured by the photon counting integral imaging system. The performance of our proposed reconstruction method is illustrated by experimental results and compared with conventional MLE using the peak signal-to-noise ratio metric. It is shown that our proposed method outperforms the conventional MLE for the photon counting 3-D integral imaging reconstruction.