Intelligent program for automatic image recognition based on compact object-fitting hierarchical image representation in terms of dynamic irregular branchy trees

Learning systems for automatic pattern recognition, which use templates, are not effective enough for non-standard objects detection. To avoid the template utilization we use an iterative object-fitting (adaptive) hierarchical segmentation which describes object fragments as the image segments computed by iterative merging (or splitting) of some initial ones. Resultant sequence of image partitions (levels) is supported by a dynamic structure of irregular trees which are introduced for optimization of computations. Due to the merging (splitting) of each segment at each iteration, a compact object-fitting hierarchical image segmentation containing a limited number of levels is generated. The accurate object recognition and a high processing speed is achieved without any increase of computer capabilities