欢迎访问林业科学,今天是

林业科学 ›› 2004, Vol. 40 ›› Issue (4): 117-123.doi: 10.11707/j.1001-7488.20040420

• 论文及研究报告 • 上一篇    下一篇

基于特征点匹配的图像拼接系统

朱正明 姜明 马小明   

  1. 北京大学数学科学学院信息科学系,北京100871北京大学环境科学中心,北京100871
  • 收稿日期:2002-10-10 修回日期:1900-01-01 出版日期:2004-07-25 发布日期:2004-07-25

An Image Mosaicing System Based on Feature Points Matching

Zhu Zhengming,Jiang Ming,Ma Xiaoming   

  1. Department of Information Science, School of Mathematical Science, Peking University Beijing100871;Center of Environmental Science, Peking University Beijing100871
  • Received:2002-10-10 Revised:1900-01-01 Online:2004-07-25 Published:2004-07-25

摘要:

对森林病虫害航空遥感监测中的图像拼接问题进行了研究。通过建立图像对应模型,按照特征点匹配的方法,并使用场景平面模型,构造了一个图像拼接系统。拼接的过程中针对图像拼接系统中特征点的提取,对应控制点的选择和对应点的搜索等问题,提出了模型参数的迭代求解方法。数据处理结果表明,使用该方法,能很好地改进了模型参数的精度,从而提高图像拼接效果

关键词: 场景平面模型, 对应关系模型, 对应控制点, 特征点, 极线约束

Abstract:

Forest insect diseases monitoring based on image mosaicing technique for aerial images were studied. Based on a general image mosaicing model and feature point correspondence technique, an image mosaicing system with planar-scene models was designed and implemented. The experiment with practical data demonstrateed the performance of the system.As for the relevant problems such as extracting feature points, selecting control points and searching correspondence points during the image mosaicing process, an iterative algorithm for estimating the model parameters was proposed. It was found that this iterative method significantly improves the accuracy of the underlying model parameters.

Key words: Planar-scene model, Image-correspondence model, Control point, Feature point, Epipolar constraint