||In this talk I will give an introduction to "particle methods", i.e. methods in which sets of samples (particles) are used to represent probability distributions. The two most important examples will be so called particle filters, which are particle equivalents of the Kalman filter, and "jet stream" which is a particle equivalent of live wire. I will also describe the advantages and disadvantages of this representation in relation to grid-based methods. Then I will describe the use of particle methods in image processing, in particular my 2007 SCIA paper Using importance sampling for Bayesian feature space filtering. The conclusion is that particle methods can be used to perform image filtering and the results are comparable to non-linear filters like the bilateral filter and median filtering.