一种图像高精度匹配方法

A High Accuracy Images Match Method

针对传统图像匹配精度不高、速度较慢的情况,为提高速度抑制噪声,提出一种高精度图像匹配方法。利用Harris提取角点进行灰度相关匹配找到粗匹配点,再利用Ransac算法得到较高精度匹配点,根据得到的匹配点求出基础矩阵。最后利用基础矩阵得到极线约束对Ransac得到的较高精度匹配点去除极少数误匹配点解算基础矩阵,利用第二次解算的基础矩阵求出高精度极线方程,并利用极线方程对Harris角点进行一维搜索匹配,找到高精度匹配点,进行仿真实验。反复实验表明,方法精度高、速度快,是一种实用的高精度图像匹配方法。

In view of that the traditional image match accuracy is not high and the speed is low, the paper puts forward a new method. This method makes use of a Harris method first to extract the characteristic points and uses the NCC algorithm to find rough match points, and then makes use of Ransae algorithm to get the refined match points. Getting the foundation matrix from the match points and calculating the epipolar constraint to reduce the wrong match points. Getting the foundation matrix again from the match points, and using the foundation matrix to get the epipolar lines, then utilizing Harris corner and making the linear dimension search to get more high accuracy images match points. Experiment result shows that this algorithm has high speed and high accuracy, and is a high accuracy image match method.