基于SIFT检测的Demons图像配准  

Based on SIFT detection of Demons

作　　者：闫德勤[1] 赵春雪[1] 刘彩凤[1] 

机构地区：[1]辽宁师范大学数学学院,辽宁大连116029

出　　处：《辽宁师范大学学报（自然科学版）》2014年第4期486-490,共5页Journal of Liaoning Normal University:Natural Science Edition

摘　　要：提出一种计算简便且配准效率高的算法.通过拟牛顿对Demons算法里的目标函数进行优化,从而得出形变向量u的方法,阐明了此方法配准的高效性,同时再结合SIFT算法对配准结果的检测,便能很直接很直观地看出配准结果的好坏,避免了判断两幅图相似性的大量计算.实验验证了所提出算法的有效性.In this paper, we propose a simple calculation and registration of high efficiency of the algo‐rithm. We optimize the objective function of Demons by quasi‐new ton algorithm, obtain the deforma‐tion vector u method, illustrates the efficiency of the method. At the same time, combining SIFT al‐gorithm to detect the registration results, can be very direct and intuitive to see the stand or fall of registration results, to avoid a large number of calculations about the similarity of two pictures. Ex‐periments show the effectiveness of the proposed algorithm.

关 键 词：Demons图像配准 SIFT尺度空间理论 关键点匹配 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]