基于改进SIFT的航拍图像快速匹配方法OACSTPCD
A Fast Method for Aerial Image Registration Based on Improved SIFT
针对传统SIFT算法存在特征描述符计算复杂,匹配时间长,错匹配较多等问题,提出了一种基于改进SIFT特征的航拍图像快速匹配方法.该方法采用基于圆形窗口的梯度方向累加值与同心圆形窗口内的灰度累加值、灰度差分值来构建18维的改进特征描述符,并在特征点匹配过程中,采用基于相关系数相似度量准则的双向匹配方法来获取初始匹配特征点对.最后,运用RANSAC算法进一步消除错配和估算仿射变换模型,并通过双线性插值法对变换后的图像进行重采样和插值.实验表明,该方法…查看全部>>
Aimming at problems that the calculation of the feature descriptor is complex, the registration time is long and wrong registrations are many in the traditional SIFT algorithm, a fast method for aerial image registration based on improved SIFT is proposed. In the method a improved feature descriptor with 18 dimension consist of 12 gradient values in a circular window, 3 cumulative gray-scale value and 3 gray-scale difference in a concentric circular window. …查看全部>>
石钊铭;耿伯英;董银文
信息技术与安全科学
图像匹配SIFT相关系数双向匹配随机抽样一致性
image registration SIFT correlation coefficient lateral matching random sample consensus
《指挥控制与仿真》 2013 (1)
106-110,5
湖北省自然科学基金项目(2011CDC023)
评论