光学技术, 2018, 44 (4): 419, 网络出版: 2018-08-30  

点对称关系耦合距离约束的图像匹配算法

Image matching algorithm based on point symmetry relation and distance constraint
作者单位
1 江苏开放大学 信息与机电工程学院, 江苏 南京 210017
2 南京信息工程大学 电子与信息工程学院, 江苏 南京 210041
摘要
为了解决当前图像匹配算法主要通过图像的区域梯度信息来进行图像匹配, 当图像存在光照变化等干扰时, 将会使得匹配图像存在较多的错误匹配点以及匹配耗时较长等不足, 提出了基于点对称关系耦合距离约束的图像匹配方法。采用FAST算法对图像特征点进行准确、快速的检测, 利用拉普拉斯极值模型进一步剔除伪特征点, 以提高算法的匹配正确度;对特征点的对称性进行计算, 利用点对称关系构造点间距度量模型, 以求取特征描述符中的特征向量, 输出特征描述符;基于SURF特征点匹配机制, 对特征点完成双向匹配约束, 完成特征点匹配。对匹配特征点进行欧氏度量, 以度量结果的比值以及均值作为依据, 构造距离约束模型, 利用距离约束模型判别错误匹配点, 优化匹配结果。实验结果显示: 与当前图像匹配算法相比较, 所提算法不仅具有较高的匹配精度以及匹配效率, 而且具有较好的鲁棒性能。
Abstract
In order to solve the defects as many mismatching points and long matching time under the conditions of illumination disturbing induced by using the region gradient information to match the image in current image matching algorithm, an image matching algorithm based on point symmetry relation and distance constraint was proposed. The FAST algorithm is used to detect the feature points accurately and quickly, and the Laplasse model is used to eliminate the false feature points for improving the matching accuracy. The symmetry of the feature points is calculated to construct the point distance measurement model for generating the feature descriptor. Based on the SURF feature point matching mechanism, the bidirectional matching constraint is implemented to realize the matching of feature points. The Euclidean measure was done for the matching feature points, and the distance constraint model was constructed based on the measurement results ratio and mean value for distinguishing error matching points to optimize feature point matching. The experimental results showed that this algorithm not only has high matching accuracy and matching efficiency, as well as robust performance compared with the current image matching algorithm in image match.

刘涛, 周先春. 点对称关系耦合距离约束的图像匹配算法[J]. 光学技术, 2018, 44(4): 419. LIU Tao, ZHOU Xianchun. Image matching algorithm based on point symmetry relation and distance constraint[J]. Optical Technique, 2018, 44(4): 419.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!