多媒体信息服务中色情图像检测算法

Pornographic image detection in multimedia messaging service

随着3G网络的不断发展,多媒体信息服务已成为移动通讯中的主要业务之一。为了向客户提供绿色健康的多媒体信息,色情画面的检测和过滤显得尤为重要。为此,本文针对3G多媒体信息服务设计了一种色情信息过滤系统。首先,提出基于混合高斯模型的肤色敏感区域提取算法,实现了对不同人种、光照下的肤色区域提取,然后结合基于主动相关反馈的Bootstrap方法进行多视角人脸检测,解决了不同光照、角度下人脸识别问题,最后,利用肤色面积、人脸在人体结构中的比例、构图规则以及敏感部位的识别结果判断图像性质。该系统已成功应用于广东移动的现网中,有效阻止了色情信息的传播。

With the rapid development of 3G networks,multimedia messaging service（MMS） has become a major form for mobile communication.To provide green and legal multimedia information,the pornographic information detection and filter for MMS is urgently demanded.For this purpose,this paper presents a pornographic information filtering system for MMS.Firstly,a skin color region extraction algorithm is proposed based on Gaussian mixtures model.It can be used to detect skin regions of people from different races and under different lighting conditions.Next,an active relevance feedback bootstrap method is adopted for multi-view face detection which solves the influence of different lighting and angle on face detection.Finally,this system employs the area of skin region,the ratio of face in body,focus of photography,and the identification results of sensitive organs to measure the sensitivity of the images.This system has been used to the MMS of Guangdong mobile company and proved that it can effectively detect pornographic images in MMS.