影像连接点均衡化高精度自动提取

发布时间：2018-05-07 01:19

本文选题：连接点提取 + SIFT　；参考：《测绘学报》2017年01期

【摘要】：针对空中三角测量影像连接点提取中存在的误匹配、点位多而分布不均和点位定位精度低等问题,提出了一种影像连接点均衡化高精度自动提取方法。首先采用分块SIFT技术进行特征的提取与匹配,并利用并查集数据结构进行特征点的高效多视追踪;然后采用提出的物方分块点位筛选算法对点位进行了相对均衡化的择优挑选;最后采用最小二乘匹配技术对得到的SIFT连接点坐标位置进行精化。选取中国嵩山遥感定标场的有人机影像和沙漠地区无人机影像作为试验数据,通过目视检查、像方反投影误差和检查点精度等3个指标进行了分析,结果表明本文方法有效克服了弱纹理和重复纹理导致的连接点提取与匹配困难,并改善了连接点分布的均匀性和提高了连接点的定位精度。
[Abstract]:In view of the problems of mismatching in the extraction of aerial triangulation images, the problem of multi location, uneven distribution and low point location accuracy, a high precision automatic extraction method for image connection point equalization is proposed. Firstly, the feature extraction and matching are carried out by block SIFT technology, and the feature points are carried out by using and searching the data structure. High efficiency and multi view tracking; then the point bit selection algorithm is used to select the relative equilibrium of the point position. Finally, the coordinates position of the SIFT connection point is refined by the least square matching technique. The human computer image and the unmanned aerial vehicle image of the Songshan remote sensing calibration field of China are selected as the test. The data are analyzed by visual inspection, such as 3 indexes such as square back projection error and inspection point accuracy. The results show that the method effectively overcomes the difficulty of extracting and matching connection points caused by weak texture and repeated texture, and improves the uniformity of connection points distribution and improves the positioning accuracy of connection points.

【作者单位】：信息工程大学地理空间信息学院;地理信息工程国家重点实验室;
【基金】：国家自然科学基金(41501482) 地理信息工程国家重点实验室开放研究基金(SKLGIE 2015-M-3-6;SKLGIE 2014-M-3-1)~~
【分类号】：P231.4

【相似文献】