A method for controlling an operating process of a printing machine uses parameters of 3-D scenes taken with only one camera. A learning phase is performed in which the object in a non-moving state and its surroundings are illuminated from different directions by different fixed-location light sources, one after another. For each illumination scene, image signals reproducing the object and its surroundings are obtained with the image recording device. The image signals are processed and an unambiguous association between object edges and object shadows in relation to a light source in each case are determined and stored. In an operating phase, the moving object and its surroundings are illuminated simultaneously by a plurality of light sources, and, from the data stored in the learning phase and the image signals obtained continuously in the operating phase, the position of the object in space is determined.