郑州轻工业学院学报(自然科学版)
鄭州輕工業學院學報(自然科學版)
정주경공업학원학보(자연과학판)
JOURNAL OF ZHENGZHOU INSTITUTE OF LIGHT INDUSTRY(NATURAL SCIENCE)
2015年
1期
63-66
,共4页
无人机航拍图像%CS-SIFT%尺度空间%准欧氏距离
無人機航拍圖像%CS-SIFT%呎度空間%準歐氏距離
무인궤항박도상%CS-SIFT%척도공간%준구씨거리
UAV aerial image%CS-SIFT%scale-space%quasi-Euclidean distance
针对无人机航拍图像对尺度变化不明显的问题,在经典SIFT特征匹配算法的基础上,提出了一种改进的CS-SIFT特征匹配算法.该算法通过建立S层金字塔,达到降低多尺度空间和减少特征点数量的目的.在特征向量的匹配中,利用准欧氏距离替代常用欧氏距离,并通过极限几何约束,消除部分错误配准点对,进一步提高特征匹配效率.Matlab仿真结果表明,改进后的算法具有较高的匹配精度和较少的匹配时间,适用于对实时性要求较高的无人机航拍系统.
針對無人機航拍圖像對呎度變化不明顯的問題,在經典SIFT特徵匹配算法的基礎上,提齣瞭一種改進的CS-SIFT特徵匹配算法.該算法通過建立S層金字塔,達到降低多呎度空間和減少特徵點數量的目的.在特徵嚮量的匹配中,利用準歐氏距離替代常用歐氏距離,併通過極限幾何約束,消除部分錯誤配準點對,進一步提高特徵匹配效率.Matlab倣真結果錶明,改進後的算法具有較高的匹配精度和較少的匹配時間,適用于對實時性要求較高的無人機航拍繫統.
침대무인궤항박도상대척도변화불명현적문제,재경전SIFT특정필배산법적기출상,제출료일충개진적CS-SIFT특정필배산법.해산법통과건립S층금자탑,체도강저다척도공간화감소특정점수량적목적.재특정향량적필배중,이용준구씨거리체대상용구씨거리,병통과겁한궤하약속,소제부분착오배준점대,진일보제고특정필배효솔.Matlab방진결과표명,개진후적산법구유교고적필배정도화교소적필배시간,괄용우대실시성요구교고적무인궤항박계통.
Aiming at the problem that UAV aerial image was not obvious to scale change,an improved CS-SIFT feature matching algorithm was presented based on the classical feature matching algorithm of SIFT (scale invariant feature transform ).The algorithm established S-layer pyramid to reduce the multi-scale space and the number of feature points.In the matching feature vectors,quasi-Euclidean distance substitu-ted commonly used Euclidean distance and geometric constraints limitd by eliminating part of the registra-tion error points to further improve the efficiency of feature matching.The simulation results using Matlab language indicated that the improved algorithm had higher matching accuracy and needed less matching time and it was quite suitable for the UAV aerial system of high real-time demand.