棒线材自动计数是一个一直没有得到很好解决的问题，为解决此问题，提出了一种基于距离变换技术的图像识别方案。该方法先利用CCD摄像机采集棒线材的端面图像；然后采用中值滤波法、二值化分割、腐蚀及膨胀运算实现棒线材图像的预处理；最后根据距离变换的思想，采用距离变换图中搜索种子点的方法实现棒线材的计数。通过对现场采集到的图像进行的试验结果表明，该方法不仅不受棒线材种类及端面形态影响，而且对端头不齐、捆扎散乱的棒线材也有较好识别效果．具有一定的推广和应用价值。

The automatic counting of wire bars and rods is an unsolved problem. To solve this problem, an image recognition approach which is based on the distance transform technique is presented in this paper. The profile images of wire bars and rods are captured by CCD cameras. In the image preprocessing the binary transform, erosion, dilation and open are conducted. The method of searching the seed spots in the distance transformed image is put forward, by which the numbers of wire bars and rods can be counted. The technique is verified by experiment that image are acquired in the factories. The counting results are not affected by the varieties and the profile shapes of wire bars and rods. This methodology is suitable for the image of wire bars and rods that are disorder and the pack are not trimness, so it is worthy of popularizing and applying.