摘要
影像匹配是无人机遥感影像拼接和三维建模的基础和关键步骤。结合不同算法的优势,本文提出一种基于特征组合与RANSAC算法的无人机遥感影像匹配方法。该匹配方法首先采用AKAZE算法检测影像的特征点,然后利用SIFT描述符描述特征向量并获取特征点的主方向,最后基于单映射变换矩阵的RANSAC算法进行精准匹配。本文对基于特征组合与RANSAC算法的匹配效果进行了试验对比分析,试验结果表明:与常用匹配方法的匹配效果相比,本文的匹配方法继承了AKAZE算法的快速匹配能力,匹配总耗时介于AKAZE算法和SIFT算法之间,约为BRISK算法匹配耗时的20%;同时,该匹配方法继承了SIFT算法的多匹配点对性能,从整体匹配效果来看,本文的匹配方法优于AKAZE、SIFT、BRISK算法。
Image matching is fundamental to remote sensing image mosaic and 3D modeling using UAV remote sensing images.A fast matching method of UAV images based on feature combination and RANSAC algorithm is proposed in this paper.The matching method first uses the AKAZE algorithm to detect the feature points of the image,and then uses the SIFT descriptor to describe the feature vectors and obtain the main direction of these feature points.Finally,the accurate matching is made by the RANSAC algorithm based on the single mapping transformation matrix.In this paper,the writer has carried out the experimental comparison analysis on the proposed matching method and algorithms. And the results were shown as follows: the matching method proposed in this paper inherits the fast matching ability of the AKAZE algorithm,with a total matching time in the middle of that consumed by the AKAZE algorithm and the SIFT algorithm,which accounts for around 20% of the time needed using the BRISK algorithm.At the same time,this matching method inherits the performance of multiple matching points of the SIFT algorithm.From the overall matching effect,the matching method in this paper is superior to the AKAZE,SIFT and BRISK algorithms.
引文
[1]关丽,丁燕杰,张辉,等.面向数字城市建设的三维建模关键技术研究与应用[J].测绘通报,2017(2):90-94.
[2]焦金龙,卢小平.顾及道路约束条件的动态目标检测与跟踪方法[J].测绘通报,2017(5):47-50.
[3]李鹏,武文波,王宗伟.基于非线性尺度空间的多源遥感影像匹配[J].测绘科学,2015,40(7):41-44.
[4]覃兵,田军,查宇飞,等.利用二值描述符的实时目标跟踪算法[J].西安电子科技大学学报,2015,42(5):168-174.
[5]王瑞瑞,冯伍法,张艳,等.一种结合相位相关法和AKAZE算法的纹理缺乏地区影像匹配技术[J].测绘通报,2017(4):49-52.
[6]牛俊伟,郝向阳,刘松林.一种改进的SIFT特征提取算法[J].测绘科学技术学报,2014,31(2):173-176.
[7]何林阳,刘晶红,李刚,等.改进BRISK特征的快速图像配准算法[J].红外与激光工程,2014,43(8):2722-2727.
[8]熊自明,万刚,闫鹤,等.改进SIFT算法的小型无人机航拍图像自动配准[J].测绘科学技术学报,2012,29(2):153-156.
[9] RUBLEE E,RABAUD V,KONOLIGE K,et al.ORB:an efficient alternative to SIFT or SURF[C]∥Proceedings of the 2001 International Conference on Computer Vision.Barcelona:IEEE,2011.
[10] ROSTEN E,PORTER R,DRUMMOND T. Faster and better:a machine learning approach to corner detection[J]. IEEE Transactions on Pattern Analysis Machine Intelligence,2010,32(1):105-119.
[11] CALONDER M,LEPETIT V,FUA P. Brief:Binary robust independent elementary features[C]∥Computer Vision ECCV 2010. Berlin:Springer,2010:778-792.
[12]陈华,邓喀中,张以文,等.结合SIFT和RANSAC算法的In SAR影像配准[J].测绘通报,2015(12):30-33.
[13]卜范秋,宋伟东,王伟玺.一种针对近景影像的匹配算法[J].测绘科学,2013,38(6):143-146.
[14]程德强,厉航,黄晓丽,等.基于改进随机抽样一致算法的视频拼接算法[J].工矿自动化,2017,43(8):50-55.
[15]邓仕雄,王晓红,刘继庚,等.基于SURF算法和极限约束的无人机影像匹配研究[J].贵州大学学报(自然科学版),2018,35(1):35-39.
[16]任克强,胡梦云,喻玲娟.基于KAZE的自适应模糊图像配准算法[J].电子测量与仪器学报,2017,31(4):559-565.
[17]贾丰蔓,康志忠,于鹏.影像同名点匹配的SIFT算法与贝叶斯抽样一致性检验[J].测绘学报,2013,42(6):877-883.