基于SVM的遥感图像半自动提取方法研究

遥感影像的半自动提取是遥感图像处理领域里的一个重要研究方向。提出一种新的半自动提取面状地物的方法,能够用随意给定的前景和背景曲线准确地提取出感兴趣的区域。算法首先提取前景和背景曲线上的像素作为分类样本,并使用SVM(Support Vector Machines)得到最初的分类效果。在此基础上,综合利用分类对象的光谱特征和形状特征信息,进行区域合并。最后,采用一种自动停止区域合并的准则确定最终合并结果。实验证明,该算法适用性强,可以在复杂多变的情况下,准确地区分出多种地物。

ON SIMI-AUTOMATICALLY EXTRACTING REMOTE SENSING IMAGE BASED ON SVM

It is an important research direction for semi-automatic extraction of the remote sensing image in the field of remote sensing image processing.This paper puts forward a new approach for semi-automatically extracting polygon features,with it the interesting region can be abstracted by the foreground and background curves which can be drawn randomly.The algorithm uses pixel values from foreground and background curves as the classifying sample and uses SVM to gain the initial classification result.On this ba...