2016/08/22 by Abderrahim Halimi, Halimi, Abderrahim, Aurora Maccarone +7 · 2 citations
Physics and Astronomy · Engineering · Medicine · #Advanced Optical Sensing Technologies #Photoacoustic and Ultrasonic Imaging #Optical Imaging and Spectroscopy Techniques
paper · pdf · doi:10.48550/arxiv.1608.06143
This paper presents two new algorithms for the joint restoration of depth and\nreflectivity (DR) images constructed from time-correlated single-photon\ncounting (TCSPC) measurements. Two extreme cases are considered: (i) a reduced\nacquisition time that leads to very low photon counts and (ii) a highly\nattenuating environment (such as a turbid medium) which makes the reflectivity\nestimation more difficult at increasing range. Adopting a Bayesian approach,\nthe Poisson distributed observations are combined with prior distributions\nabout the parameters of interest, to build the joint posterior distribution.\nMore precisely, two Markov random field (MRF) priors enforcing spatial\ncorrelations are assigned to the DR images. Under some justified assumptions,\nthe restoration problem (regularized likelihood) reduces to a convex\nformulation with respect to each of the parameters of interest. This problem is\nfirst solved using an adaptive Markov chain Monte Carlo (MCMC) algorithm that\napproximates the minimum mean square parameter estimators. This algorithm is\nfully automatic since it adjusts the parameters of the MRFs by maximum marginal\nlikelihood estimation. However, the MCMC-based algorithm exhibits a relatively\nlong computational time. The second algorithm deals with this issue and is\nbased on a coordinate descent algorithm. Results on single-photon depth data\nfrom laboratory based underwater measurements demonstrate the benefit of the\nproposed strategy that improves the quality of the estimated DR images.\n