Since the invention of metal foams they became an attractive material for different industrial and engineering applications. The physical backgrounds of metal foaming, relations between the microstructure of metal foams and their macroscopical properties, and controllable foaming, however, are still disputed topics. In this work we develop image processing methods for the quantitative investigation of coalescence processes, for the characterization of material flows during foaming, and for the recognition of basic structural elements such as lamella and pores. First the methods are evaluated on synthetic projection image sequences. Further, they are applied to the radiographic sequences of real metal foaming. The backgrounds of the methods are a new coalescence measure and developed technique for the detection of coalescence events, a novel combination of image processing steps for the recognition of round bubbles, and an established tensor approach for the estimation of velocity vectors.