Extracting the structure of 2-D structural objects using combination of self-organizing maps and Karhunen-Loeve transform based on recursive bifurcation

Abstract

Identifying the characteristics, structures, and features of a given set of data is the most essential step prior to the classification, recognition, and data interpretation process. Extracting the identified features must preserve data composition configuration. Any irrelevant features will be excluded after the extraction process. This study concerns the extraction of the actual 2-dimensional geometrical element frame from a set of 2-dimensional structural objects. Our approach is based on Self-Organizing Mapping and Karhunen-Loeve transformation that does not require predetermined number of clusters in advance as oppose to other approaches.