摘要
为了控制网络空间的色情传播,利用浏览器助手对象可以在浏览器的地址空间内引入处理代码,构建对Web页面色情图片的过滤器。色情图片在色度空间YUV和YIQ中分布具有一定特性,根据U与V的夹角和I的阈值进行肤色检测,生成掩码图像,然后进行纹理处理,提高了准确性。在掩码图像的基础上在原图像中提取特征值,进行图像的分类(色情图片或正常图片),决定对Web页面的处理,解决了Web页面色情图片过滤这一难点而又紧迫解决的问题。
BHO(Browser Helper Object) can bring management code into the address space of browser. A filter of erotic images based on BHO was built for web pages. Distribution of erotic images has some characteristics in YUV chroma-space and YIQ chroma-space. The scopes of angle (of\% U\% and\% V\%) and I can detect the skin area of the images and produce the mask-code images. The texture processing of the images can improve the veracity of the skin detection. Characters drown from the images based on the mask images can classify the images (erotic images or normal images), they decide how to deal with the web pages. The difficult and urgent problem of the web-page erotic image filter is resolved.
出处
《吉林大学学报(信息科学版)》
CAS
2004年第6期631-637,共7页
Journal of Jilin University(Information Science Edition)