Rapid Target Modeling Through GET Inheritance Mechanism

The objective of this project is to develop a Genetic Evolution Target (GET) modeling methodology for rapid target model development and validation from limited initial sensory and technical data. Developed methodology will allow for an expansion of the initial signature database and extensive testing of ATR systems over a variety of realistic signatures and situations not seen so far. The methodology utilizes a power of genetic inheritance to generate new signatures of a given target from a limited set of initial signatures and a reduced 3-D geometric model of the target. Intial signatures are represented as strings of morphological processes applied to some set of predefined shapes. These strings, when executed, constitute given shapes of signature elements. The population of strings obtained from a starting set of signatures is subject to genetic evolution. Crossover and mutation operations are applied during each evolutionary cycle to generate new signatures. New signatures in the population are validated according to closeness to a reduced 3-d geometric model of a target supplied by the user. Only converging signatures are kept in the population. After a number of evolution cycles, signatures in the current population inherit significant resemblance from the initial subset of signatures and are characterized by introduced additional changes.