Object recognition method based on clustering analysis of fuzzy situation

The article highlights the aspects determining various conditions that complicate obtaining and formalizing the initial actual information about the object being monitored or localized at a certain point in time. Taking into account the above aspects, it is recommended to use fuzzy information as the basis for the methods of object recognition. To develop this direction, it is proposed to use so-called algebra-scales and appropriate methods of scaling and clustering to obtain and convert fuzzy information. In order to generalize this information, the article uses the concept of fuzzy situation. The current and typical situations are isolated. Their aggregates define (characterize) both the observed image of the object, and its possible typical image, which corresponds to specific circumstances. In order to recognize the image of an object more quickly, it is recommended to use cluster analysis, which allows to combine typical situations into clusters with certain properties, and to trace the movement of the current situation inside the specified clusters using the fuzzy inclusion or equality relations. This, to a significant degree, makes it possible to increase the efficiency of deciding on an appropriate response to the behavior of the observed object. It is proposed to develop and program such solutions in advance. The article contains analytical expressions of algorithms necessary for implementing the method of object recognition and provides an example of clustering of typical situations using matrices.