《计算机应用研究》|Application Research of Computers

基于三角网下的仿射不变几何约束的图像匹配算法研究

Study on affine invariant geometric constrained image matching algorithm based on triangle mesh

免费全文下载 (已被下载 次)  
获取PDF全文
作者 王恒,王怀柱,刘艳青
机构 宁夏大学 a.计算机网络与信息管理中心;b.信息工程学院,银川 750021
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2017)08-2528-05
DOI 10.3969/j.issn.1001-3695.2017.08.065
摘要 针对当前图像匹配方法的鲁棒性差、误配率较高及效率较低等不足,提出了基于三角网下的仿射不变几何约束的图像匹配算法。在尺度空间上通过Hessian矩阵对特征点进行检测,利用子块的三角特征与对角特征SURF(speeded up robust features)机制进行改进,用于生成新的特征描述子,并通过定义阈值评估策略对图像特征点进行匹配,从而生成了初始匹配点;然后,引入Delaunay三角网,对初始匹配点进行聚类,以获取匹配三角形,将三角形以外的无效特征点剔除;最后,引入仿射不变几何约束,对匹配三角形进行细化,通过细化的匹配三角形获取最终的匹配特征点,有效剔除误配点,进一步提高配准精度。仿真结果表明,与当前图像匹配算法相比,所提算法具有更好的鲁棒性,且其具有更佳的匹配精度与效率,有效剔除了误配点。
关键词 图像匹配;Delaunay三角网;仿射不变几何约束;SURF机制;Hessian矩阵
基金项目 国家自然科学基金资助项目(61562069)
宁夏高等学校科学技术研究基金资助项目(NXGXZR201107)
宁夏自然科学基金资助项目(61563043)
本文URL http://www.arocmag.com/article/01-2017-08-065.html
英文标题 Study on affine invariant geometric constrained image matching algorithm based on triangle mesh
作者英文名 Wang Heng, Wang Huaizhu, Liu Yanqing
机构英文名 a.Network&InfomationManagementCenter,b.CollegeofInformationEngineering,NingxiaUniversity,Yinchuan750021,China
英文摘要 In order to solve these drawbacks such as the poor robustness and large matching error and low efficiency of the current image matching method in image matching, this paper proposed an affine invariant geometric constrained image matching algorithm based on triangle mesh. First, it used the Hessian matrix determinant in the scale space to detect feature points, and used the triangular and diagonal feature of the block to generate feature descriptor by generation method of SURF (speeded up robust features). Then it defined the threshold evaluation strategy to generate initial matching points for matching feature points. It introduced the Delaunay triangle mesh to cluster the initial matching points and obtained the matching triangle and removed invalid feature points beyond the triangle. Finally, it used the affine invariant geometric constraint to refine the matching triangles, and obtained the final matching feature points by the refined matched triangles, effectively eliminated the wrong matching points, and further improved the registration accuracy. The simulation results show that, compared with the current image matching algorithm, this algorithm has better robustness and better matching accuracy and efficiency, effectively eliminates the wrong matching points.
英文关键词 image matching; Delaunay triangle mesh; affine invariant geometric constraint; SURF; Hessian matrix
参考文献 查看稿件参考文献
  [1] Lindeberg T. Image matching using generalized scale-space interest points[J] . Journal of Mathematical Imaging and Vision, 2015, 52(1):3-36.
[2] Luce J, Gray J, Hoggarth M A. Medical image registration using the Fourier transform[J] . International Journal of Medical Physics, Clinical Engineering and Radiation Oncology, 2014, 3(1):49-55.
[3] 陈凯, 吕文阁, 谢庆华. 基于竞选算法的灰度图像匹配研究[J] . 四川理工学院学报:自然科学版, 2014, 27(5):69-71.
[4] Sedaghat A, Ebadi H. Very high resolution image matching based on local features and K-means clustering[J] . Photogrammetric Record, 2015, 30(15):166-168.
[5] 许佳佳, 张叶, 张赫. 基于改进Harris-SIFT算子的快速图像配准算法[J] . 电子测量与仪器学报, 2015, 29(1):48-53.
[6] Zhao Dong, Yang Yan, Ji Zhihang. Rapid multimodality registration based on MM-SURF[J] . Neurocomputing, 2014, 131(5):87-97.
[7] Huang Liqin, Chen Caigan, ShenHenghua. Adaptive registration algorithm of color images based on SURF[J] . Measurement, 2015, 66(1):118-124.
[8] 罗楠, 孙权森, 陈强. 结合SURF特征点与DAISY描述符的图像匹配算法[J] . 计算机科学, 2014, 41(11):286-300.
[9] Abdelrahman H, Elsalamony M. Comparing proposed signature with SURF in object detection process[J] . IETE Journal of Research, 2015, 61(5):466-474.
[10] 吴一全, 沈毅, 陶飞翔. Delaunay 三角形剖分约束下的图像配准算法[J] . 遥感学报, 2014, 18(3):618-629.
[11] 张东兴, 祝明波, 李相平. Delaunay 三角形剖分约束下的图像配准算法[J] . 信号处理, 2013, 29(12):1644-1649.
[12] Dongjoe S, Tardi T. Clique descriptor of affine invariant regions for robust wide baseline image matching[J] . The Journal of the Pattern Recognition Society, 2014, 43(10):3261-3272.
[13] Dou Jianfang, Qin Qin, Tu Zimei. Improved SVD matching based on Delaunay triangulation and belief propagation[J] . International Journal for Light and Electron Optics, 2015, 126(24):4861-4867.
[14] 常卢峰. 基于三角形相似性的图像匹配算法研究[J] . 科技广场, 2010, 4(3):99-101.
[15] 徐衍鲁, 马燕, 李顺宝. 结合颜色不变量的SIFT和形状上下文图像匹配算法[J] . 计算机科学, 2014, 41(11A):144-146.
[16] 李婕, 邓德祥, 石文轩. 一种抗视角变换的SURF匹配算法[J] . 计算机应用研究, 2012, 29(9):3536-3539.
[17] Tanvir H M, Teng S W. Multimodal image registration technique based on improved local feature descriptors[J] . Journal of Electronic Imaging, 2013, 24(1):1-17.
[18] 龚卫国, 张旋, 李正浩. 基于改进局部敏感散列算法的图像配准[J] . 光学精密工程, 2011, 19(6):1375-1382.
收稿日期 2016/5/19
修回日期 2016/7/9
页码 2528-2532
中图分类号 TP391.41
文献标志码 A