Abstract: Based on the analysis of the traditional edge-adaptive algorithm，the reasons leading to the low vertical resolution in the traditional edge-adaptive algorithm are discussed.A modified edge dependent de-interlacing algorithm that combines the time-domain filtering，spatial filtering and median filtering，is introduced，the pixel value is calculated with the bidirectional information of forward frame and the backward frame，then a median filter is followed.After the simulation of software and the analysis of theory，the algorithm is proved as an effective edge dependent de-interlacing algorithm，which can improve the vertical resolution properly and ensure the sharpness of edge.

Key words: de-interlacing, median filtering, edge adaptive, video processing

摘要： 在分析了传统边沿自适应去隔行算法的原理及优缺点的基础上，详细地对传统边沿法的低垂直清晰度造成的原因进行了探讨和解释，从理论上对其不足进行弥补改进，提出了一种结合时域滤波、空间滤波和中值滤波的改进的边沿检测去隔行算法。利用待插值点前后场的信息以及当前场的信息对检测出的边沿待插值像素进行双向插值后，再进行中值滤波。软件仿真结果和理论分析说明：改进的算法，引入较多的原始图像信息进行去隔行处理，有效地提高了垂直清晰度，同时保证了边界的锐利，具有良好的视觉效果。

关键词: 去隔行, 中值滤波, 边沿自适应, 视频图像处理