基于图像处理的枪弹头痕迹自动匹配方法

Automatic Matching Method of Bullet Traces Based on Image Processing

针对子弹自动自动匹配问题，首先利用子弹表面为圆柱面的假设，对采集得到的数据进行误差校正，然后将圆柱面上的深度数据展开到平面上，用经典的中值滤波和平滑算法去除噪声，良好的数据预处理算法对后续的特征提取和子弹匹配起到了重要作用．一方面将z轴数据映射到二维图像上，提出了基于统计的鲁棒的互相关性系数准则，另一方面，对三维数据进行可视化操作，显示划痕条数、划痕宽度等宏观特征，二者结合起来对子弹进行匹配．实验表明，方法取得了良好的效果，具有高达80％以上的识别率．

This paper assumes that the surface of bullet is cylindrical for matching bullet automatically. Firstly, the acquisition errors is corrected; secondary, the depth data of the cylindrical surface is mapped to the plane; thirdly, noise is eliminated with classical median filter and smoothing algorithm. Good preprocessing algorithm plays an important role in feature extraction and bullet matching later. On one hand, z-axis data is mapped to two-dimensional images, and a robust correlation criterion is proposed, on the other, the three-dimensional data is displayed with scientific visualization, showing the macroscopic characteristics such as Number of scratch, scratch width and so on. Experiment shows that this method has achieved good results, with more than 80% recognition rate