一种基于多分类器融合的双色红外成像目标识别方法

针对双色红外成像系统中的自动目标识别问题,提出了一种基于多分类器融合的红外目标识别方法.该方法首先提取目标的形状特征和面貌特征,并设计多个基于不同特征的分类器对目标进行分类;然后对各个分类器的目标分类结果进行决策级融合处理,并采用所提出的决策规则对多分类器融合分类结果进行处理得到最终的目标识别结果.该方法充分利用了目标在多传感器图像中的多种分类特征信息,提高了系统的目标识别效率和精确性.实验结果证实了该方法的有效性.

A Kind of Method for the Two Color IR Imaging Target Recognition Based on Multi-Classifiers Fusion

Aim at the problem of ATR for the two color IR imaging system,proposed a kind of method for IR target recognition based on multi-classifiers fusion.The algorithm firstly extracted the shape features and appearance-based features of target and devised multiple classifiers based on the different features to classify the target;Then,fused the target classification results of multiple classifiers; Finally,deal with the multi-classifiers fusion classification results using devised decision rules to get the system's target recognition re- suits.The method made good use of the multiple kinds of classification feature information of target in multi-sensors images,improved the performance and aecuracy of target recognition.Experimental results demonstrated the effectiveness of the proposed method.