A-few-photons active imaging method under high-flux ambient light

Fast active imaging using a few photons under high ambient light flux is challenging in applications. Based on the local spatial correlation, existing methods have reduced the requirement for photon counts in histograms. Here, we introduce an adaptive clustering method into active imaging for strong ambient light scenes, which focuses on reconstructing accurate depth images with the elapsed time in second order. We theoretically demonstrate the feasibility of estimating optimally the incident light flux using only 5 photons per pixel. Furthermore, we provide an outstanding superpixel strategy to cluster the pixels with high spatial and intensity global correlations by the adaptive clustering radius based on the power transformation of the estimator. The simulations and experiments indicate that the proposed method outperforms the other advanced methods in terms of adaptivity, depth accuracy, and elapsed time, which significantly benefits its application.