###
工程科学与技术:2014,46(5):96-103
←前一篇   |   后一篇→
本文二维码信息
码上扫一扫!
基于边缘特征的虚实图像精确配准
(四川大学计算机学院视觉合成图形图像技术国防重点学科实验室)
PreciseRegistrationofVirtual andRealImagebyEdgeFeature
(NationalKeyLab.ofFundamentalSci.onSyntheticVision,CollegeofComputerSci.,SichuanUniv.)
摘要
图/表
参考文献
相似文献
本文已被:浏览 2087次   下载 0
投稿时间:2014-02-28    修订日期:2014-06-08
中文摘要: 为了将3D模型的虚拟图像与可见光图像(实图)配准融合以增强信息的互补性,将虚实融合问题变成经典的多模态图像精确配准问题后,继而提出了一种基于边缘信息全局最优匹配的图像精确配准算法。该算法首先将虚实图像通过惯导装置粗配准;然后基于相位一致性模型提取3D模型图像与可见光图像视觉相似的边缘结构;最后采用模拟退火算法搜索求解在边缘配准度最大对应的空间变换模型的最优参数值。4组实验结果表明,本文算法能够充分利用图像间的相近似的边缘结构和模拟退火算法得到配准变换参数,有效地实现多种多模态图像的精确配准。
Abstract:In order to enhance the complementary advantages by registration and fusion of virtual and real image, a precise registration algorithm by edge feature and global optimization was proposed and the problem was transformed to a classical multi-modal image registration problem. At first, initial registration was achieved by inertial navigation system(INS) instruments. Then the 3D model and visual image edge structure similar with human vision was extracted by the edge algorithm based on phase congruency model. Finally the global optimal parameter with respect to maximal registration value of space transforming model was calculated by simulated annealing algorithm. 4 groups of experiments result showed that this registration method can make full use of the similar edge structure between 2 images; and simulated annealing algorithm can achieve the better parameter for registration transform. The experiments result also indicated that these multi-modal pairs are registered with enough accuracy.
文章编号:201400198     中图分类号:    文献标志码:
基金项目:国家“863”计划资助项目(2013AA013802)
作者简介:
引用文本:
李新胜,刘宇.基于边缘特征的虚实图像精确配准[J].工程科学与技术,2014,46(5):96-103.
Li Xinsheng,Liu Yu.PreciseRegistrationofVirtual andRealImagebyEdgeFeature[J].Advanced Engineering Sciences,2014,46(5):96-103.