基于SIFT和非参贝叶斯的高分辨率遥感影像地物识别算法
发布时间：2019-01-05 03:06
【摘要】：地物识别是遥感图像处理领域中的一个重要问题。随着遥感技术的发展,高分辨率遥感影像中携带有大量相似的具有尺度不变特征的地物,传统的地物识别方法难以适应这一发展,亟需对其进行改进。针对高分遥感影像,在SIFT(Scale-invariant Feature Transform)算法的基础上进行改进并得出一种快速精准的地物识别算法DBSIFT(Double Backward SIFT),实现了相似地物多对一的模式识别。DBSIFT在原算法的基础上构造了二重差金字塔,利用DP(Dirichlet Process)识别出相似地物并对其进行分割。在几何与算数关系上,选取9个指标对分割精度进行评价。实验中,使用该方法得到的地物能够被准确识别,且分割效果良好,说明了该算法的有效性。
[Abstract]:Ground object recognition is an important problem in the field of remote sensing image processing. With the development of remote sensing technology, high resolution remote sensing images carry a large number of similar features with scale invariant features, the traditional method of ground object recognition is difficult to adapt to this development, it is urgent to improve it. Based on SIFT (Scale-invariant Feature Transform) algorithm), a fast and accurate ground object recognition algorithm, DBSIFT (Double Backward SIFT), is developed for high score remote sensing images. DBSIFT constructs the pyramid of double difference based on the original algorithm and uses DP (Dirichlet Process) to recognize and segment the similar ground objects. In the relation between geometry and arithmetic, nine indexes are selected to evaluate the segmentation accuracy. In the experiment, the ground objects obtained by this method can be accurately identified and the segmentation effect is good, which shows the effectiveness of the algorithm.
【作者单位】： 山西大学计算机与信息技术学院;山西大学计算智能与中文信息处理教育部重点实验室;
【基金】：国家自然科学基金资助项目(41101440) 山西省青年科技基金资助项目(2012021015-1)资助
【分类号】：TP751
本文编号：2401178
[Abstract]:Ground object recognition is an important problem in the field of remote sensing image processing. With the development of remote sensing technology, high resolution remote sensing images carry a large number of similar features with scale invariant features, the traditional method of ground object recognition is difficult to adapt to this development, it is urgent to improve it. Based on SIFT (Scale-invariant Feature Transform) algorithm), a fast and accurate ground object recognition algorithm, DBSIFT (Double Backward SIFT), is developed for high score remote sensing images. DBSIFT constructs the pyramid of double difference based on the original algorithm and uses DP (Dirichlet Process) to recognize and segment the similar ground objects. In the relation between geometry and arithmetic, nine indexes are selected to evaluate the segmentation accuracy. In the experiment, the ground objects obtained by this method can be accurately identified and the segmentation effect is good, which shows the effectiveness of the algorithm.
【作者单位】： 山西大学计算机与信息技术学院;山西大学计算智能与中文信息处理教育部重点实验室;
【基金】：国家自然科学基金资助项目(41101440) 山西省青年科技基金资助项目(2012021015-1)资助
【分类号】：TP751
【参考文献】
相关期刊论文 前3条
1 葛盼盼;陈强;顾一禾;;基于Harris角点和SURF特征的遥感图像匹配算法[J];计算机应用研究;2014年07期
2 陈梦婷;闫冬梅;王刚;;基于Harris角点和SIFT描述符的高分辨率遥感影像匹配算法[J];中国图象图形学报;2012年11期
3 张林;刘辉;;Dirichlet过程混合模型的聚类算法[J];中国矿业大学学报;2012年01期
【相似文献】
相关期刊论文 前3条
1 龚建周;陈健飞;刘彦随;;基于E0-1 Hyperion影像地物识别与分类不同方法的效果比较[J];应用基础与工程科学学报;2013年03期
2 刘行华;胡宝新;张渊智;;地物理解的现状、设想与展望[J];遥感信息;1992年04期
3 ;[J];;年期
相关硕士学位论文 前1条
1 李益;基于光谱/空间联合特征的遥感影像地物提取技术研究[D];解放军信息工程大学;2012年
，本文编号：2401178
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2401178.html
