Abstract: It becomes increasingly important to locate degraded video（video with defects） with the wide application of digital media.For simple form of digital video defects，such as black frame，red，blue and green frame，detection algorithms are proposed.In these algorithms，some typical features，such as mean，Absolute Difference（AD），are used.For the complex video flaw，exactly mosaic，a fast and effective detection algorithm is proposed based on analysis of existing approaches.Two tricks are adopted to speed up the detection：A new template matching strategy and the preprocessing of the edge image.All above defect detection algorithms are implemented based on the OpenCV library，which make the development of the system pleasant and easy.Experiments are carried out and the results show that the proposed algorithm has good performance.

Key words: digital video defect, computer vision, OpenCV

摘要： 随着视频应用范围的扩展，在视频中定位出视频缺陷变得越来越重要。对于简单的视频缺陷，比如黑场，红、绿、蓝场以及静帧的缺陷，提出了相应的检测算法，用到的特征包括绝对差等这类简单的特征。对于复杂的视频缺陷，如马赛克缺陷，在分析现有算法的基础上，提出了一个快速有效的检测算法，采用了两种策略以提高检测速度：新的模板匹配及对边缘图像的预处理。所有以上检测算法都基于OpenCV库开发实现，使得整个算法的实现更快捷。进行了仿真实验，结果验证了所提算法的有效性。

关键词: 数字视频缺陷, 计算机视觉, OpenCV