摘要
在建筑三维遥感图像优化的研究中,利用遥感图像合成远距离的三维建筑是一项重要的工作。在利用传统方法进行大型建筑遥感三维建模时,由于遥感数据包含大量的噪声,造成匹配过程出现大量错误,建筑三维数据之间的拓扑关系很难实现。提出基于多角度图像的大型建筑三维建模办法,在提取匹配立体像对中的同名点时,利用了Harris算法与SIFT特征匹配算法相结合的方法,对多角度图像目标及场景的三维信息提取时,采用遥感成像理论的同名点对匹配方法和传感器参数模型,利用泰勒级数将传感器参数成像模型线性化,通过误差迭代的方式对参数进行精细标定。仿真结果表明,该方法能够有效提高对大型建筑目标及场景三维信息提取的精度,同时也增强了系统的鲁棒性。
A three-dimensional modeling method for large building is presented based on multi-angle image, the Harris algorithm combined with SIFT feature matching algorithm is used to extract the matched stereo same point, the corresponding point matching method and the model of sensor parameters based on the remote sensing imaging theory are used to extract the multi-angle image target and scene 3d information. The imaging model of sensor parameters is linearized by using Taylor＇s series, and the parameters are calibrated by means of error iteration. Simulation results show that the method can effectively improve the extraction accuracy of large construction target and scene 3d informa- tion, and enhance the robustness of the system.
出处
《计算机仿真》
CSCD
北大核心
2015年第9期254-258,共5页
Computer Simulation
基金
黑龙江省自然基金项目(F201439)
黑龙江省教育厅科学技术研究项目(12541899)
校级项目(2014123)
关键词
大型建筑
特征匹配算法
遥感成像理论
同名点对匹配方法
Large construction
Feature matching algorithm
Remote sensing imaging theory
Corresponding pointmatching method
