采用径向Tchebichef矩不变量的飞机型号识别

Aircraft type recognition using radial-Tchebichef moment invariants

针对几何矩非正交性对目标描述的不足以及连续正交矩在处理数字图像方面存在离散化误差的缺陷,为了提高识别精度,提出了一种利用离散正交的Tchebichef矩结合全局特征和局部特征的飞机型号识别方法。首先,根据几何矩和Tchebichef矩之间的关系,利用归一化几何中心矩、圆谐函数得到径向Tchebichef矩的旋转、尺度和平移(RST)不变量;然后,利用径向Tchebichef矩提取飞机目标的局部和全局特征构成特征向量;最后,利用Matlab构造了四类飞机的样本集,采用支持向量机(SVM)作为分类器识别测试样本飞机型号,分析了几何矩、Zernike矩和本文方法在识别精度上的差异以及训练样本集大小对识别精度的影响。实验结果表明,本文提出的算法提高了识别精度,并且在训练样本集较小时仍能获得90%以上的识别精度。

According to the shortage of geometric moment nonorthogonality in obje ct description and the defect of continuous orthogonal moments in processing digital image with discretization er ror,in order to improve the recognition accuracy,this paper proposes a new aircraft type identification method with dis crete orthogonal Tchebichef moments combined with global and local features.Firs t of all,according to the relationship between the geometric moments and Tchebichef moments,the normalized geometric center moment and circular harmonic function are used to get rotation, scale and translation （RST） invariant radial-Tchebichef moments;Then,global and local features of aircraft target are extracted to form feature vector by the radial Tchebichef moments;Finally,four types of sample s et of the planes are constructed through Matlab program,the support vector machine （SVM） is used as classifier to identif y the aircraft type of test sample set which consists of the whole sample set except the training set,and the differ ences among geometric moment,Zernike moment and the proposed method are compared in recognition accuracy and the effect of t he training sample set size on the identification accuracy.Experimental results show that the proposed algorithm improves the rec ognition accuracy,and the recognition accuracy is still greater than 90% when the training sample set is small.