针对大量视频图像中出现的各种文字信息，本文提出了一种基于离散余弦变换（DCT）的文字提取算法。该方法首先将图像分割为等大小基本块，然后对各小块提取DCT特征。在此基础上，利用图像对比度，设计了一种动态阈值分割方法，可将文字信息和背景信息进行分离。然后依据最小外接矩形算法，获得初始文字检测结果。最终使用Voronoi Diagram算法对初始区域进行合并得到最终文字区域检测结果。算法可以快速而精确定位文字所对应的区域，并且能适用于各种背景条件下的视频图像。

To help users navigate the libraries of video, algorithms that automatically index video based on the content are needed. In this paper, we present a DCT based approach to detect texts and captions from the videos. The use of these features is in a flexible manner thus can be adapted to different applications. Language independence is an important advantage of the proposed method. Experiments are conducted on a large volume of real video shots. Solutions are proposed for each of these problems and compared with the existing work found in the literature.