一种结合纹理特征的极化SAR图像SVM分类方法

Integrating Texture Features in Polarimetric SAR Image Classification Using SVM

目前,基于目标分解特征的传统极化SAR图像分类方法,通常不能满足使用者对精度的需求,因此有必要进一步提高其分类精度。纹理特征是提高地物辨识度的重要特征,将目标分解与其结合可以有效增加特征向量在分类中的作用,改善极化SAR分类过程中精度低的问题,因此,本文提出一种结合纹理特征的极化SAR图像分类方法。实验结果表明,在不同的特征向量中,结合纹理特征后,各类地物的分类精度以及总体分类精度均有不同程度提高。

At present, the traditional polarimetric SAR image classification method based on target decomposition fea- ture usually can not meet the usefs need for accuracy, so it is necessary to improve classification accuracy. Texture fea- tures are important to improve objects＇ recognition, integrating target decomposition is helpful to increase the feature vector effect in classification, and improve the low precision of polarimetric SAR classification problems. Therefore, this paper presented a method of polarimetric SAR image classification based on texture feature. The experimental re- sults showed that this method could improve the classification accuracy of all kinds of objects and overall accuracy in different feature vector integrating texture features.