摘要
针对传统RANSAC算法在特征点提纯方面效率不高、迭代计算复杂等缺点,提出先利用拓扑约束进行特征点提纯,得到初始匹配点集,再通过RANSAC原理进行特征点精确提纯,最后通过最小二乘法利用精确匹配点求解单应矩阵进行图像配准。实验结果表明:拓扑约束提纯算法计算效率高,能有效提高RANSAC算法的正确匹配率和时间效率,可得到更多更稳定的匹配点,提高图像配准的精度。
To improve the efficiency and complexity of traditional RANSAC algorithm in purifying matching points,a strategy by using topology restraint is put forward.Firstly utilize topology restraint between feature points to get initial matching pairs and then precisely purify by using RANSAC algorithm.Then holography matrix is obtained by Least Squares with the exact matching pairs for image registration.Experimental results show that,compared with traditional RANSAC algorithm,the proposed method performs well in efficiency while improves the robust matching rate and more stable matching pairs can be acquired to improve the precision of image registration.
出处
《测绘与空间地理信息》
2016年第9期158-160,共3页
Geomatics & Spatial Information Technology
关键词
特征点提纯
拓扑约束
误匹配
匹配率
purifying matching points
topological constraints
mismatching
matching rate
