Integrating color features in polarimetric SAR image classification
-
摘要: 为了提出一种颜色特征与极化特征相结合的极化SAR图像分类方法,首先,通过极化目标分解得到极化特征向量;然后,采用最佳指数模型方法生成极化SAR的假彩色合成图像,并提取颜色特征向量;最后,将这2种特征组成综合特征向量,利用SVM方法进行分类.利用RadarSat-2的PolSAR数据进行了SAR图像分类实验,并对分类结果进行定性和定量比较分析.实验结果表明,颜色特征的加入能有效提高极化SAR图像的分类精度.Abstract: This paper presents a method for combining the color feature and target decomposition characteristics so as to study the classification of polarimetric SAR.It makes up decomposition feature vector by polarimetric target decomposition and then, through the pseudo color enhancement method, obtains the false color image of polarimetric SAR data representation;after that, it extracts color histogram from the pseudo color images to make up the color feature vector, thus providing additional information for further land classification.Classification experiments were performed at different feature vectors by using RadarSat-2 polarimetric SAR image.In addition, the quantitative and qualitative comparison analysis was conducted with classification results.The experimental results show that the addition of the color feature can effectively improve the classification accuracy of polarimetric SAR images.
-
Key words:
- PolSAR image classification /
- pseudo color enhancement /
- color feature /
- feature vector
-
-
计量
- 文章访问数: 404
- PDF下载数: 29
- 施引文献: 0
下载: