基于权重策略的不良图像识别  

Recognition on porn image based on weight strategy

作　　者：周立前[1] 胡柳[1] 李瑞[1] 黄丽君[1] 胡盛龙[1] 文志强[1] 

机构地区：[1]湖南工业大学计算机与通信学院,湖南株洲412007

出　　处：《中南大学学报（自然科学版）》2013年第11期4561-4565,共5页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(61170102);湖南省自然科学基金资助项目(13JJ3109);湖南省教育厅重点项目(13A004)

摘　　要：针对网络上不良图像泛滥问题,在人脸检测、肤色检测、图片纹理等分析方法的基础上,提出基于权重的不良图像识别的新方法。首先,选取合适的颜色空间;然后,对肤色区域的纹理信息进行提取;最后,采取特征权重策略,应用支持向量机方法对网络上搜集的不良图像和正常图片的相应特征进行训练与分类。研究结果表明:该方法对不良图像识别正确率达88.85%,正常图片识别正确率为86.70%,能应用于实际软件系统中对不良图像进行过滤。For the problem that the porn images on the network spread unchecked, a new method based on weight strategy was proposed for the recognition of porn image based on the face detection, skin-color detection, image texture analysis and other methods. The appropriate color space was firstly selected. Then the texture information from skin region was extracted. Finally the corresponding characteristics ofporn image and normal picture collected from network were trained and classified by the feature weight strategy and the method of support vector machine （SVM）.The results show that the accuracy rate of this method is 88.85% for the recognition of porn image and 86.70% for normal picture, and the method can be applied to filter porn image in actual software system.

关 键 词：权重策略 不良图像 肤色检测 支持向量机 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]