摘要
在大量的航空图像中如何自动准确而且快速地识别出飞机型号,对于战地指挥员有着重要的意义。提出一种将傅里叶变换和奇异值分解相结合的航空机型自动并行识别方法。首先,对航空图像进行傅里叶变换,得到其具有位移不变特性的振幅谱表征特征;其次,对图像进行奇异值分解,用类估计基空间作为同类图像奇异值分解时对应的基空间;最后,在集群系统下用最近邻法从航片中自动识别出机型。
It is important for commander recognizing plane type accurately and fast from mass of aviation images.A type of aviation images recognition method based on Fourier transform and Singular Value Decomposition(SVD) is introduced.Firstly,aviation images are processed by a 2D Fourier transform that has some effective characters such as a linear transform and invariant against spatial transform.Secondly, SVD is used for subtracting the character of the amplitude image,and using base-space of class to estimate as congener images' standard space.Finally,using the traditional Nearest Neighbor Classifier(NNC) based on PC cluster to tell the unknown type from aviation images.
出处
《计算机工程与应用》
CSCD
北大核心
2008年第34期185-187,共3页
Computer Engineering and Applications
基金
重大科技攻关项目(No.072SGZS38042)
关键词
傅里叶变换
奇异值分解
集群
消息传递接口
机型识别
Fourier transform
Singular Value Decomposition(SVD)
cluster
Message Passing Interface(MPI)
type recognition