论文部分内容阅读
数字视频数据中的文本能提供重要的语义信息，用具有良好时频局部特性和变尺度特性的小波分析方法，提取图像中的尺度一空域特征，经神经网络分类器进行分类后，自动检测视频数据中的文本信息，为基于内容的视频检索提供索引．实验表明该检测方法效果良好。
The text in digital video data can provide important semantic information. With the help of the wavelet analysis method which has the characteristics of good time-frequency local feature and variable scale, the scale-space features in the image are extracted. After classification by the neural network classifier, the video is automatically detected Textual information in the data, providing indexing for content-based video retrieval. Experiments show that the detection method works well.