自动视频分类中的数据预处理

提出了一套视频数据采集、特征提取及数据预处理方法。介绍了视频数据的采集方式，并提出了一套可计算的视频特征属性，可以反映人们对不同的视频风格的典型特征的理解。给出了如何将这些提取的视频属性数据处理成适合于决策树分类、分类关联规则以及序列模式等数据挖掘技术适用的数据格式。

Data Preprocess in Automatic Video Classification

In order to facilitate automatic video classification based on data mining technique, this paper proposs a data preprocess method including video data collection and feature extraction. The method of video data collection is introduced. It presents a set of computational attributes, which can reflect human's understanding of the typical features of various video genres. It also describes how to convert these video attributes into a new data form that is fit for data mining such as decision tree, class association rule and sequential pattern.