摘要
为解决无人机遥感拼接图像易受地貌因素影响,产生形变或重影等问题,提出一种改进SPHP算法。用SURF算法和Harris算法相结合快速提取特征点,将得到的特征点用KNN算法粗匹配,用PROSAC算法进行精匹配,引入权重系数,计算图像重叠区域的空间变换模型,将该模型代替SPHP算法原有空间模型,降低图像重叠区域的重影,使拼接后的图像产生更小的形变。实验结果表明,该算法可以有效去除拼接图像的重影,生成更好的拼接结果。
An improved SPHP algorithm was proposed to solve the problem that UAV remote sensing mosaic images are easily affected by geomorphological factors,resulting in deformation or ghost.Feature points were extracted by combining SURF algorithm and Harris algorithm.The obtained feature points were roughly matched using KNN algorithm.They were matched accurately using PROSAC algorithm.The weight coefficient was introduced to calculate the spatial transformation model of the image overlap region.This model replaced the original spatial model of SPHP algorithm.The ghost of the image overlap area was reduced,and the stitched image had smaller deformation.Experimental results show that the proposed algorithm can remove the ghost of the stitched image effectively and generate better stitching results.
作者
王红君
刘一鸣
岳有军
赵辉
WANG Hong-jun;LIU Yi-ming;YUE You-jun;ZHAO Hui(School of Electrical and Electronic Engineering,Tianjin University of Technology,Tianjin 300384,China;College of Engineering and Technology,Tianjin Agricultural University,Tianjin 300384,China)
出处
《计算机工程与设计》
北大核心
2020年第3期783-788,共6页
Computer Engineering and Design
基金
天津市科技计划基金项目(17YFCZZC00330、18YTZCNC01120、17ZXYENC00080)。