一种基于区域加权的飞机识别方法  

Recognition of Aircrafts Based on Weighted Area

作　　者：李渝[1] 吴增印[1] 

机构地区：[1]西安空间无线电技术研究所,西安710071

出　　处：《计算机科学》2013年第11A期350-353,共4页Computer Science

摘　　要：为了区分不同种类的飞机,根据飞机结构的特殊性,提出了一种基于区域加权的飞机识别方法。首先利用Gabor变换的多尺度多方向性提取飞机的机头、机翼、机尾3个有效区域,分别对这3个区域提取特征并识别,然后根据不同区域对飞机全局特征的贡献为3个区域分配权重,最后结合权重将不同区域识别结果进行融合得到最终的飞机类型。实验结果表明在相同数量级识别时间的条件下,本文飞机识别方法比传统的支持向量机、神经网络等方法有更高的识别率,同时有较强的抗遮挡效果,是一种有效的飞机目标识别方法。An aircraft identification method based on weighted area was proposed to distinguish different aircrafts by means of the difference of their structures. Firstly, obtain the head, wing and tail of the aircraft using the multi-scale and multi-direction characters of Gabor transform, extract the features of the three areas and recognise them respectively,then distribute the weight for the three regional according to the different regional contribution to the aircrafts＇ global features, finally combine with the weight of the recognition results in different regions to get the final aircraft type. The results show that this method has a higher recognition rate than traditional support vector machine and neural network methods under the same order of magnitude recognition time conditions and a strong anti-overlap effect, so it is an effective method of aircraft target recognition in this paper.

关 键 词：GABOR变换 特征提取 权重 抗遮挡 识别率 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]