摘要
针对传统图像拼接算法存在拼接速度慢、图像拼接有色差等问题,提出了一种基于ORB-GMS-SPHP算法的大视场多图像拼接方法。该方法首先利用高斯函数构建尺度空间,将高斯尺度空间划分为多个网格,在每个网格内借助FAST算法提取尺度空间特征点,使用BRIEF算法提取描述符并匹配,得到更加均匀分布的特征点;然后使用运动网格统计算法筛选匹配点;最后采用SPHP算法融合图像重叠区域,从而得到完整的拼接图像。将改进的ORB-GMS-SPHP算法与现有的传统特征点匹配算法在特征点匹配精度和特征点匹配速度进行对比与评价,验证了该方法特征点匹配速度快、精度高,并且可以保留更多的正确匹配点的特点。将该拼接方法与传统拼接方法在拼接速度、图像配准均方误差RMSE以及视觉主观判断拼接色差进行对比与评价,验证了该拼接方法具有较快的拼接速度、更高的拼接精度和无明显色差。该方法在2736像素×3648像素图像中,特征点匹配时间降低至6.463 s,图像配准精度RMSE降低至3.87。实验证明该方法特征点匹配速度快、精度高,且拼接精度高、无明显色差。
Aiming at the problems of slow mosaic speed and poor color of image mosaic in traditional image mosaic algorithms,a large field of view multi-image mosaic method based on ORB-GMS-SPHP(oriented FAST and rotated BRIEF_grid-based motion statistics shape-preserving half-projective)algorithm is proposed.Firstly,the Gaussian function is used to construct the scale space,and the Gaussian scale space is divided into several grids.In each grid,the FAST algorithm is used to extract the feature points of the scale space,and the BRIEF algorithm is used to extract the descriptors and match them,so as to obtain the feature points with more uniform distribution.Then the motion grid statistical algorithm is used to screen the matching points.Finally,the method of SPHP is adopted to fuse the overlapping area of the image,so as to get a complete mosaic image.The comparison and evaluation of the improved ORB-GMS-SPHP algorithm and the existing traditional feature point matching algorithm in feature point matching accuracy and speed verify that the proposed method has fast matching speed and high accuracy,and can retain more correct matching points.The proposed mosaic method is compared and evaluated with the traditional mosaic method in mosaicing speed,image registration mean square error RMSE and visual subjective judgment mosaicing color difference.It is verified that the proposed mosaicing method has faster mosaicing speed,higher mosaicing accuracy and no obvious color difference.In the 2736×3648 pixel image,the feature point matching time is reduced to 6.463 s,and the image registration accuracy RMSE is reduced to 3.87.The experimental results show that the method has fast matching speed,high accuracy and no obvious color difference.
作者
何佳华
申冲
唐军
刘俊
HE Jiahua;SHEN Chong;TANG Jun;LIU Jun(National Key Laboratory of Dynamic Testing Technology,North University of China,Taiyuan 030051,China;School of Instrument and Electronics,North University of China,Taiyuan 030051,China)
出处
《导航定位与授时》
CSCD
2023年第2期108-116,共9页
Navigation Positioning and Timing
基金
国家自然科学基金创新研究群体(51821003)
国家自然科学基金优秀青年基金(51922009)
国家自然科学基金面上项目(61973281)
山西省重点研发计划项目(202003D111003)
山西省优秀青年培育项目(202103021222011)。
