基于改进SURF算法的360°全景图像拼接方法  

360°Panoramic Image Stitching Method Based on Improved SURF Algorithm

作　　者：冀福全 姚莉[2] 汪先超 JI Fuquan;YAO Li;WANG Xianchao(School of Information Engineering,Chaohu University,Hefei 238024,China;School of Computer Science and Engineering,Southeast University,Nanjing 211189,China)

机构地区：[1]巢湖学院信息工程学院,安徽合肥238024 [2]东南大学计算机科学与工程学院,江苏南京211189

出　　处：《宿州学院学报》2023年第6期16-22,共7页Journal of Suzhou University

基　　金：江苏省产业前瞻与共性关键技术项目(BE2018119);巢湖学院人才启动经费项目(KYQD2013)。

摘　　要：针对虚拟现实、旅游景区、装修展示等领域日益增长的宽视野图像需求,提出一种水平360°全景图像拼接方法,该方法基于相邻视角两图像拼接,对原始图像添加柱面映射预处理,通过改进的SURF算法进行特征点提取,使用K最邻近算法和RANSAC算法进行特征点匹配和过滤,在图像融合过程中结合了线性权重融合算法和非线性权重融合算法。拼接过程借助多线程并发处理,使多组相邻图像的拼接操作同步进行。实验结果表明,该方法能够拼接得到水平360°全景图像,图像变形程度低、过渡自然,具有较好的稳定性和拼接效果,同时相较于传统SURF算法,拼接速率具有较大提升。Aiming at the increasing demand for wide-field images in virtual reality,tourist attractions,decoration and display,a horizontal 360°panoramic image Mosaic method is proposed.Based on the stitching of two images from adjacent perspectives,the cylindrical mapping preprocessing is added to the original image,and the feature points are extracted through the improved SURF algorithm.The improved SURF algorithm was used to extract feature points,and the K-nearest algorithm and RANSAC algorithm were used to match and filter feature points.In the process of image fusion,linear weight fusion algorithm and nonlinear weight fusion algorithm are combined.By means of multi-thread concurrent processing,the stitching operation of multiple groups of adjacent images can be carried out synchronously.Experimental results show that this method can obtain horizontal 360°panoramic images with low deformation and natural transition,and has good stability and stitching effect.Meanwhile,compared with traditional SURF algorithm,the stitching rate is significantly improved.

关 键 词：快速鲁棒特征算子 图像拼接 全景图像 图像配准 图像融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]