基于HCA与KAZE的铁路路基GPR图像配准算法

Railway Ballast GPR Image Registration Algorithm Based on HCA and KAZE

针对铁路路基探地雷达(GPR)检测需要较高精度与时效性的要求,提出一种基于直方图曲率分析(HCA)与KAZE特征的图像配准算法。通过HCA进行阈值分割,提取图像高能量区域,以节约无效区域的配准时间。运用KAZE算法提取图像中的特征点,并根据快速近似最近邻搜索算法进行粗匹配。使用随机抽样一致性算法过滤匹配点对,优化特征匹配过程。实验结果表明,该算法对于存在病害差异、增益差异、地物差异的铁路路基多时相GPR图像均取得较好的配准效果,且配准精度比KAZE算法、ORB算法、SIFT算法明显提高,配准效率比KAZE算法提升8%以上。

Aiming at the problem that railway ballast Ground Penetrating Radar（GPR） detection need high timeliness and accuracy,an image registration algorithm based on Histogram Curvature Analysis （HCA） and KAZE features is proposed. It extracts high energy region of image using HCA threshold segmentation,in order to save time for registration useless region. KAZE features in GPR images are extracted and features matching is optimized through Fast Library for Approximate Nearest Neighbors （FLANN） algorithm and Random Sample Consensus （RANSAC） algorithm. Experimental results show that the proposed algorithm can get good registration results for GPR images including disease, grayscale and structure difference. And it gains higher accuracy than original KAZE algorithm, ORB algorithm and SIFT algorithm and more than 8% higher efficiency than original KAZE algorithm.