一种基于高层语义的文字图像过滤模型

A text image filtering model based on high-level semantics

针对包含印刷体汉字的文字图像提出一种过滤模型,该模型把图像中每一个文字看成一个对象,通过相关技术手段从图像中分割出对象并对这些对象进行识别,将敏感词汇组成部分的对象标记为敏感对象。同时,建立了语义词典,通过组合将敏感对象组合起来并与词典中的词汇相匹配,从而达到识别敏感词的目的,以此为基础可以实现对含有敏感词的图像的过滤。

A filtering model of text image which contain printed Chinese characters is presented in this paper. In this model, each character in the image is regarded as an object. The appropriate techniques is adopted to implement image segementation and object recognition. If an object is part of a sensitive word,this object is labeled as sensitive object. A semantic dictionary was created. The sensitive objects are grouped and then coupled with the words in this dictionary,so as to recognize the sensitive word and realize the filtering of the images which contain sensitive words.