Performance evaluation for automatic target recognition based on cloud theory
Fang Wang, Yanpeng Li, Xiang Li · 2008
In the recent years, as the development of automatic target recognition (ATR) technology, the performance evaluation for it becomes more and more important. However, the existent evaluation methods always have shortcomings such as needing too many samples or needing artificial intervene to determine some parameters like weights. In order to solve these problems, we proposed a new performance evaluation method for ATR System based on cloud theory. In this method, we use numerical value features which are contained in the recognition result samples to denote the performance of the ATR algorithm and then use the weight clouds to replace the index weights in the fuzzy comprehensive evaluation. Therefore we can do the evaluation without artificial intervene and with much less number of samples compared with other existent ones. At last, we compared our evaluation method with the general fuzzy comprehensive evaluation through the simulations with real measurements and the simulation proved the advantages above.