There are different ways to identify sites for future excavation in archaeology. One of these methods is based on magnetic properties of hidden underground objects and is carried out by means of magnetometric survey. In the pre-selected points the magnetic induction module of geomagnetic field is measured. Then the data is preprocessed (for example, changes in the Earth’s total magnetic field are taken into account). After that, the researchers build a map of the area and estimate the probability that there are objects of archaeological value in the selected area. Based on this map the decision is taken whether it is worth to conduct excavations. However, both useful anomalies caused by the presence of archaeological sites and anomalies of a random nature can be simultaneously presented on the map. Unfortunately, sometimes it is quite difficult for a human to detect with one’s eyes what anomalies are useful in the picture. In this article a process of automatic allocation of segments of archaeological objects is described. For this purpose specific smoothing and adaptive threshold binarization methods as well as clustering algorithms are considered.