Digital field fragment classification and object clusterization using the fast search method

The method of fast classification of digital field fragments and clusterization of analogous objects has been developed using the stack algorithm of the information transmission theory. The algorithm operation speed is studied as a function of the resemblance measure. The computational complexity is evaluated. The method is tried in the recognition of planar patterns. It has also been used for processing a two-dimensional digital field comprised of a set of time sequences. The algorithm allows the real-time building of the dictionary tree for a large number of time series and clusterization of input vectors for a particular resemblance measure.