Target recognition for the two color IR imaging system based on the multi-classifiers decision level fusion

摘要

Aim at the problem of Automatic Target Recognition (ATR) for the two color IR imaging system, presented a method for the IR dual band image target recognition based on multi-classifiers decision level fusion. This method firstly inputted all kinds of feature vectors of the target image into these relevant classifiers respectively to get the likelihood ratio of the target image fall into every class according to the outputs of these classifiers; Then, fused the outputs of these classifiers using Transformable Belief Model(TBM) theory to get the decision probability distribution of the target belong to these different classes for the whole system; Finally, analyzed and decided the decision probability distribution according to the decision rule to get the final recognition result for the target image. These experimental results at the end of this paper showed the effectiveness of the presented method.