通过对现有板坯表面裂纹检测方法的分析，提出了一种基于图像特征的板坯表面裂纹检测方法，阐述了运用该方法的系统架构及具体实现过程。该方法以网络摄像机和工控机为系统的核心硬件平台，由网络摄像机采集的数字图像经以太网发送给工控机。在工控机端用VS2008编写软件实现对板坯上表面及侧面状态的实时监视以及对每个板坯的录像，以人工在线标记缺陷为主，自动识别算法为辅，检测出板坯表面的缺陷；工控机端软件还可以对录下的历史板坯录像进行回放，重新进行缺陷的检测。自动识别算法采用与图像亮度信息无关的相位一致性算法。

By means of analysis of detecting methods for surface cracks of the existing slab, a detection method based on image feature to surface crack of slab was put forward. The system architecture and the specific implementation process were described. The method took network cameras and IPC as the core of the system′s hardware platform. The digital image acquired by the network camera was sent via Ethernet to IPC.A software developed in IPC by VS2008 realizes the real-time monitoring to the state of the upper surface and the side of the slab, and the video of each slab was record. The system supplemented by automatic recognition algorithm mainly artificial line marked defects detect surface defects of slab. The software in IPC can also replay the recorded slab video to re-defect defects. Phase consistency algorithm for automatic recognition algorithm is used, which has nothing to do with the brightness information of image.