首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于Levenberg-Marquardt算法的图像拼接
引用本文:伏燕军,杨坤涛,邹文栋,何兴道.基于Levenberg-Marquardt算法的图像拼接[J].四川激光,2007,28(5):46-48.
作者姓名:伏燕军  杨坤涛  邹文栋  何兴道
作者单位:华中科技大学光电子科学与工程学院,华中科技大学光电子科学与工程学院,南昌航空工业学院无损检测技术教育部重点实验室,南昌航空工业学院无损检测技术教育部重点实验室 湖北武汉430074,南昌航空工业学院无损检测技术教育部重点实验室,江西南昌330063,湖北武汉430074,江西南昌330063,江西南昌330063
摘    要:Levenberg-Marquardt(简称LM)算法是高斯-牛顿法和梯度下降法的结合,既有高斯牛顿法的快速收敛性,也有梯度下降法的全局搜索特性.将LM算法应用于图像拼接过程中变换矩阵的参数自动优化中,通过对多个对应点的自动优化后,可以较快得到变换矩阵,从而实现图像拼接.在LM算法中,以对应点距离的平方和作为目标函数,给出此方法详细的实现过程与结果.与以灰度差平方和为目标函数相比,避免了提前采用插值算法,从而求得的M变换矩阵更精确.结果表明,该方法实现图像拼接的精度高,速度快.

关 键 词:图像拼接  Levenberg-Marquardt算法  参数优化  变换矩阵  插值  插值算法  图像拼接  marquardt  algorithm  based  mosaic  速度  精度  过程与结果  度差平方  方法  目标函数  距离  参数自动优化  对应点  变换矩阵  应用  特性  搜索  快速收敛性  高斯牛顿法
文章编号:0253-2743(2007)05-0046-03
修稿时间:2007-03-22

Image mosaic based on levenberg- marquardt algorithm
FU Yan-jun,YANG Kun-tao,ZOU Wen-dong,HE Xing-dao.Image mosaic based on levenberg- marquardt algorithm[J].Laser Journal,2007,28(5):46-48.
Authors:FU Yan-jun  YANG Kun-tao  ZOU Wen-dong  HE Xing-dao
Institution:1. School of Opto-electronic Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, China; 2. Laboratory of Nondestructive Test ( Ministry of Education
Abstract:The Levenberg-Marquardt Algorithm is the combination of the gradient decent Algorithm with the Gauss-Newton algorithm.It has the ability of both fast convergence and global search.The LM Algorithm is used in parametric automatic optimization of the transformation matrix in Image Mosaic.After parametric automatic optimization of corresponding points,the transformation matrix is quickly got,and then the image mosaic is realized.In the LM Algorithm,the distance sum of squares of the corresponding points is taken as the target function.The detail realization and result of the method is given.Compared with the gray level difference sums of squares as the aim function.The Interpolation Algorithm is avoided advance,then the M transformation matrix is more accurately.The result indicates that the method of image mosaic has high precision and fast speed.
Keywords:image mosaic  levenberg-Marquardt algorithm  parametric optimization  transformation matrix  interpolation
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号