Fuzzy Run Theory and Its Application in Performance Evaluation of Automatic Target Recognition

The fuzzy comprehensive evaluation has played an important role in performance evaluation, unfortunately, however, it may lose information of the system being evaluated. We establish the theory of fuzzy run which can give more objective conclusion in fuzzy comprehensive evaluation, as can help the latter in building more practical evaluation models. There are three components brought to fuzzy mathematics in this theory: (i) the "bottom run" to show the difference of current output and the lowest limitation in n characteristic dimensions; (ii) the "limitation of the bottom run" to assure the variance of the former quantity; and (iii) the "run between dimensions" indicate the difference between n dimensions. . Subsequently, aiming at performance evaluation of Automatic Target Recognition (PE-ATR, ATR), a general idea is developed; the evaluation indexes and their measurement method are presented; then, a performance evaluation model is formed. Furthermore, simulation is carried out with test data, and the results show the practicability of this newly proposed facility. This new method is beneficial for evaluating an ATR system in: (i) it can take many different kinds of performance characters into account in the evaluation course; (ii) it is easy to be configured; it is an open model, as users can easily change the evaluation set and the comment set.