摘要： 作为一种基本图像类型,文本图像在电子商务等方面得到了广泛应用。针对图像数据库中文本图像识别与分类的应用需求,依据文本图像数据与连续色调图像的总体灰度分布差异,该文提出了一种基于图像信息度量(Picture Information Measure,PIM)的图像特征和基于该特征的文本图像分类方法。约2G、800幅网络图像数据库的分析和实验表明,图像的PIM特征可以显著区分文本图像和连续色调图像,识别和分类效果好。

关键词: 图像检索, 文本图像, 图像信息度量, 识别分类

Abstract: As a basic type of images, document images have found a more and more application in e-office, e-commerce,etc.To meet the needs of document image identification in image datebase, this paper introduces the image feature of PIM and provides the classification argirithm based on it. The experiment on the 2G image database shows that the image characters extended from PIM performs effectively in classifying document images from those images with continously-distributed hues.

中图分类号: 

• TP391.41