【作者】 罗继强； 许廷发； 潘腾； 徐超；

【Author】 Luo Jiqiang;Xu Tingfa;Pan Teng;Xu Chao;Beijing Institute of Technology,School of Optics and Photonics;Beijing Institute of Spacecraft System Engineering;

【机构】 北京理工大学光电学院； 北京空间飞行器总体设计部；

【摘要】 随着卫星遥感图像分辨率的提升,感兴趣目标在图像中所占尺寸越来越小,从地面复杂背景中快速提取出感兴趣的目标依然具有挑战性。本文提出了一种基于局部学习方法的直推支持向量机的遥感图像目标提取方法。该方法将半监督学习方法和局部学习方法相结合,能够有效利用已知目标样本的直推支持向量机,实现对小样本遥感目标识别的分类器构建。实验结果表明,该方法在高分辨率卫星遥感图像上的海上目标提取上比其他方法有更好的性能。更多还原

【Abstract】 With the enhancement of satellite remote sensing image resolution,the target in remote sensing image is small,and it is still challenging to quickly extract the target of interest from the satellite remote sensing images from the complex ground background.In this paper,one novel target extraction method of remote sensing image based on local learning method and transductive support vector machine is proposed.This method combines the semi supervised learning method with the local learning method,and can effectively use the transductive support vector machine(SVM)of the known target samples to realize the construction of the classifier for the small sample remote sensing target recognition.Experimental results show that the proposed method has better performance than other methods in marine target extraction on high resolution satellite remote sensing images.更多还原