摘要
提出一种基于内容的敏感图像判别模型以识别不良图像。该模型结合颜色特征提取、皮肤检测和人脸检测等技术提取图像的视觉特征,并通过Bayes分类器完成底层特征到高层语义的映射。提出一种基于链表的连通域探测算法,结合肤色和纹理构成皮肤检测模型,实验表明该模型对皮肤检测有较准确的结果。
Develops an erotic image identification model based on content.The identification model synthesizes techniques of color feature extraction,skin recognition and face recognition to express the visual characters of images.It completes the transformation from visual characters to semantic characters of image by Bayes algorithm.In addition,proposes a pixel labeled algorithm based on chain table of connecting area,which composes skin recognition model with skin and texture recognition.The result of local experiment shows that this identification model can achieve high recognition rate.：
出处
《现代计算机》
2010年第9期47-50,共4页
Modern Computer
基金
武汉科技大学创新基金项目
关键词
颜色空间
视觉特征
语义识别
贝叶斯算法
Color Space
Visual Character
Semantic Recognition
Bayes Algorithm
