针对雷达高分辨率距离像(HRRP)识别中因特殊样本和分类器误判而出现的错误分类问题,提出了一种基于自适应类别权重的多分类器决策融合识别方法。该方法结合K-最近邻思想,利用最近邻和相似度准则挑选与测试样本对应的训练样本集,构造混淆矩阵自适应完成分类器置信度的计算和筛选,最终获得目标各类别权重,输出分类结果。基于实测数据的研究结果表明,相较于以上任意单个分类器和传统决策融合方案,文中提出的融合识别方法识别率有明显提高,并且随着噪声的增大,该方法的优势愈加突显。

Aiming to resolving the problem of misclassification resulting from the special samples and the misjudgment of classifier in radar high resolution range profiles (HRRP), a multi-classifier decision fusion based on adaptive-classification-weight method is proposed in this paper. In this method, the K-nearest neighbors and the similarity rule are combined to select the training sample set corresponding to the testing sample, then the confusion matrix is constructed to complete the confidence calculation and screening of classifier adaptively, and finally the classification weights of targets can be obtained and classification result is outputted. Experimental results based on the measured data show that the recognition rate of the proposed method in this paper is better than any of the above classifiers and traditional decision fusions. With the increment of noise interference, the advantage of the proposed method becomes more obvious.

TN957. 5

国家自然科学基金资助项目(61071163、61271327、61471191);航空基金资助项目(20152052026);江苏省研究生培养创新工程(SJLX16_0109);研究生创新基地(实验室)开放基金资助项目(kfjj20160401);中央高校基本科研业务费专项资金资助