屈玳辉,谢益武.基于APAP模型的大视差图像拼接算法[J].测控技术,2021,40(1):34-39
基于APAP模型的大视差图像拼接算法
Large Vision Difference Image Splice Algorithm Based on APAP Model
  
DOI:10.19708/j.ckjs.2021.01.005
中文关键词:  ORB算法  MLDB描述符;向量场一致性算法;APAP模型
英文关键词:ORB algorithm  MLDB descriptor  vector field consistency algorithm  APAP model
基金项目:
作者单位
屈玳辉 大连海事大学 信息科学技术学院 
谢益武 大连海事大学 信息科学技术学院 
摘要点击次数: 411
全文下载次数: 115
中文摘要:
      针对待拼接图像具有大视差时,重叠区域会出现重影和拼接效率较低等问题,提出一种基于改进ORB算法的大视差图拼接模型。首先构建尺度空间,借助FAST算法提取特征点并建立主方向,再对特征采样区域建立MLDB描述符,然后采用向量场一致性算法筛选内点,最后通过APAP模型完成配准拼接。实验表明,与ORB算法相比,改进ORB算法的鲁棒性更好,与SIFT算法和AKAZE算法相比,匹配速度更快;本算法拼接与GlobalHomography算法、传统APAP模型和AANAP模型相比,效率更高、拼接效果更好,符合视觉审美。
英文摘要:
      In order to solve the problems of low efficiency and double shadow in overlapping region when the image to be spliced has large disparity of sight,an improved ORB algorithm is proposed.The scale space is constructed,the feature points are extracted and the main direction is established with the help of the FAST algorithm,the MLDB descriptor is established for the feature sampling area,the vector field consistency algorithm is used to screen the inner points,and the registration splicing is completed by APAP model.The experiment shows that the improved ORB algorithm has better robustness than the ORB algorithm,and the matching speed is faster compared with the SIFT algorithm and the AKAZE algorithm.This algorithm has higher efficiency,better spliced effect than GlobalHomography algorithm,traditional APAP model and AANAP model,and accords with visual aesthetic.
查看全文  查看/发表评论  下载PDF阅读器
关闭