Many countries were the target of air strikes during the Second World War. The aftermath of such attacks is felt until today, as numerous unexploded bombs or duds still exist in the ground. Typically, such areas are documented in so-called impact maps, which are based on detected bomb craters. This paper proposes a stochastic approach to automatically detect bomb craters in aerial wartime images that were taken during World War II. In this work, one aspect we investigate is the type of object model for the crater: we compare circles with ellipses. The respective models are embedded in the probabilistic framework of marked point processes. By means of stochastic sampling the most likely configuration of objects within the scene is determined. Each configuration is evaluated using an energy function which describes the conformity with a predefined model. High gradient magnitudes along the border of the object are favoured and overlapping objects are penalized. In addition, a term that requires the grey values inside the object to be homogeneous is investigated. Reversible Jump Markov Chain Monte Carlo sampling in combination with simulated annealing provides the global optimum of the energy function. Afterwards, a probability map is generated from the automatic detections via kernel density estimation. By setting a threshold, areas around the detections are classified as contaminated or uncontaminated sites, respectively, which results in an impact map. Our results, based on 22 aerial wartime images, show the general potential of the method for the automated detection of bomb craters and the subsequent automatic generation of an impact map. © Authors 2019.