Finn Lindgren, Matematisk statistik i Lund: Flame
  reconstruction by Markov Chain Monte Carlo simulation

Abstract:

In combustion engineering it is of interest to determine the
properties of flame propagation, such as the flame volume and
area at a given time after ignition. To this end, several
orthogonal image pairs, one for each cycle, of the flame in a
spark ignition engine have been obtained, as well as
measurements of several other quantities.

The flame is modelled as a deformed ellipsoid, using Markov
Random Fields. The flame model is combined with a model for
the two orthogonal views to obtain sample flame reconstructions
using MCMC. The reconstruction algorithm works without help
from an operator, and is able to estimate the areas of the
cooling surfaces by subtracting the spark plug and cylinder
walls from the flames. A complete MATLAB implementation of the
algorithm has been made.