基于机器视觉的齿轮图像特征匹配优化

Gear Image Feature Matching Optimization Based on Machine Vision

为了提高视觉系统中齿轮图像匹配的准确率,提出了一种金字塔光流法与ORB(oriented fast and rotated brief)算法结合的齿轮图像匹配优化方法。该方法在图片进行灰度化和高斯模糊等预处理操作后,用ORB算法得到图像的特征点,用金字塔光流法跟踪图片间的特征点,暴力匹配之后通过汉明距离和哈曼顿距离进行筛选,最后以改进的RANSAC(random sample consensus)算法优化匹配结果。实验结果表明,该算法在齿轮图像匹配过程中能够减少错误的匹配点,筛选出具有代表性的匹配点,减少误匹配率。

To improve the accuracy of gear image matching in the vision system,a gear image matching method combining the pyramid optical flow method and the oriented fast and rotated brief(ORB)algorithm is proposed.This method uses the ORB algorithm to obtain the feature points of the image after the image is preprocessed by grayscale and Gaussian filtering,and then uses the pyramid optical flow method to track the feature points between the images.After brute force matching,the Hamming distance and the Harmanton distance are passed to filter.Finally,the improved random sample consensus(RANSAC)algorithm is used to optimize the matching results.Experimental results show that the algorithm can reduce false matching points in the gear image matching process,screen out representative matching points,and reduce the false matching rate.