基于不变矩和支持向量机理论的船舰目标识别

基于支持向量机(Support Vector Machine,SVM)理论和不变矩(Invariant Moments)理论,提出一种船舰目标识别方法。首先,对图像进行预处理,将彩色图像转化为灰度图像;然后利用Hu不变矩来提取图像的七个不变矩特征;最后,选用支持向量机作为分类器,并将计算出的图像的七个矩特征作为支持向量机的输入对支持向量机进行训练和测试。实验证明,将不变矩特征提取方法与SVM相结合用于模式识别,可以得到很高的分类效率和准确率。

Recognition of ship by support vector machine and Invariant Moments

The paper gives a method of ship recognition which based on the Support Vector Machine and Invariant Moments.Firstly,make pretreatment for the image,that is,transform the multicolor image to gray image;then compute seven invariant moment characters by Hu Invariant Moments;At last,we chose Support Vector Machine as the classifier,and make the seven invariant moments characters as the input of Support Vector Machine to training and predict the classifier.The experiment indicated that Pattern Recognition by the SVM and Invariant Moments has higher class efficiency and veracity.