中文摘要：

针对钢轨磨耗动态测量中激光光条中心快速精确提取的问题，提出一种卡尔曼滤波和Hessian矩阵相结合的激光光条中心快速提取方法。首先，利用卡尔曼滤波实时预测钢轨磨耗动态测量中激光光条在图像中所在区域；然后，在预测的激光光条区域内，逐行搜索图像灰度最大点，将该灰度最大点作为激光光条图像中心的初始位置，在激光光条图像中心初始位置处利用Hessian矩阵计算得到光条中心的亚像素图像坐标；最终实现在激光光条区域内光条亚像素图像中心的快速提取。该方法显著减少了搜索区域及高斯卷积的数目，提高了激光光条中心提取的鲁棒性及速度。实验结果表明，在保证激光光条提取精度的前提下，每帧提取时间可达到1．6ms。

英文摘要：

A center-extracted algorithm by combining the Kalman filter with the Hessian matrix was proposed for the implementation of real-time stripe processing in rail wear dynamic measurement. Firstly, the image region of a laser stripe in rail wear dynamic measurement was predicted with the Kalman filter method. Then,the points with max gray were searched in the predicted region of laser stripe line by line, and they were considered as the initial positions of laser stripe image center. Subpixel positions of laser stripe image center were then extracted through the calculation of Hessian matrix in the initial positions of laser stripe image center. Finally, the sub-pixel image center of laser stripe was obtained in the region of stripe. The proposed algorithm reduces the searching area and the number of Gaussian convolutions greatly, and improves the robustness and speed in the extraction of laser stripe centre in the rail wear dynamic measurement. Experimental results show that it takes 1.6 ms to process every frame, and the extraction accuracy of laser stripe is also guaranteed.