Scattering Tomography by Monte Carlo Voting

In this paper, we propose a new tomography method called Monte Carlo Voting to visualize the crosssectional surfaces of scattering media. Since the incident light repeats a collision with small particles in scattering media, there are many light paths. To trace them, we use the Monte Carlo ray tracing algorithm and compute the attenuation ratio that represents the likelihood of hidden objects for each light path. The distribution of hidden objects is estimated by voting the attenuation ratio along the light path. Monte Carlo Voting statistically solves the tomography problem by combining Monte Carlo ray tracing and the voting process. Experiments using simulated and real images show that the distribution of hidden objects can be roughly estimated even when strong scatterings are observed.