论文部分内容阅读
面向对象分类过程,首先对图像进行分割得到对象,然后将对象进行分类,分割效果直接影响最终分类精度.针对这一问题,提出一种改进的全极化合成孔径雷达(SAR)影像面向对象分类方法,在分类时首先通过计算各对象内部像元类别比例对对象进行判断,若所有类别比例均没有达到某个阈值,则认为此对象存在分割偏差,对其进行基于像元的分类,反之则进行面向对象分类,最后整合像元级和对象级分类结果.分类算法采用改进分类器动态选择法(ICDS)对差异性较大的3个基分类器Wishart、核-KNN和Wishart-KNN进行决策级融合.以AIRSAR,EMISAR的全极化SAR影像为数据进行分类实验.结果表明:改进算法充分利用了对象级和像素级分类的优点,从而得到高精度的分类结果,该算法具有良好的应用前景.
In the process of object-oriented classification, firstly the image is segmented to get the object, then the object is classified and the segmentation result directly affects the final classification accuracy.In view of this problem, an improved object-oriented classification of all-polarimetric synthetic aperture radar (SAR) In the classification, firstly, the object is judged by calculating the ratio of the pixel types within each object. If all of the classifications do not reach a certain threshold, the object is considered to have segmentation deviation, and pixel-based classification is performed on the object, and vice versa And finally integrate the pixel-level and object-level classification results.The classification algorithm adopts the improved classification dynamic selection method (ICDS) to make decisions on Wishart, KNN and Wishart-KNN Level fusion.The experimental results show that the improved algorithm makes full use of the advantages of object-level and pixel-level classification to obtain high-precision classification results, and the algorithm has good application prospect.