计算机工程 ›› 2011, Vol. 37 ›› Issue (21): 202-204.doi: 10.3969/j.issn.1000-3428.2011.21.069

• 图形图像处理 • 上一篇    下一篇

一种改进的敏感图像过滤方法

陆 蓓,陈法叶,姚金良   

  1. (杭州电子科技大学计算机学院,杭州 310018)
  • 收稿日期:2011-05-27 出版日期:2011-11-05 发布日期:2011-11-05
  • 作者简介:陆 蓓(1960-),女,教授,主研方向:图形图像处理,人工智能;陈法叶,硕士研究生;姚金良,博士
  • 基金项目:
    浙江省科技厅重大专项基金资助项目(2010C11049); 浙江省自然科学基金资助项目(Y1080883)

Improved Sensitive Image Filtering Method

LU Bei, CHEN Fa-ye, YAO Jin-liang   

  1. (Institute of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou 310018, China)
  • Received:2011-05-27 Online:2011-11-05 Published:2011-11-05

摘要: 针对现有敏感图像过滤方法误检率较高的问题,提出一种结合肤色检测和方向梯度直方图(HOG)人体检测的敏感图像过滤方法。采用HOG特征提取人体目标的特征集,运用支持向量机训练人体检测模型,检验图像中是否存在人体,并结合肤色检测算法判别该图像是否为敏感图像。实验结果表明,该方法能有效检测复杂背景条件下的敏感图像,其精确度为90.2%、查全率为86.3%、误检率为3.5%。

关键词: 敏感图像, 肤色检测, 人体检测, 支持向量机, 方向梯度直方图

Abstract: Aiming at the shortage that existing sensitive image filtering method has higher error rate, this paper presents a sensitive image filtering method by combining skin color detection with human detection by Histogram of Gradient(HOG). Extracting features of human bodies by HOG feature, using a detection model trained by Support Vector Machine(SVM) to find out the human body, and then using the skin color detection algorithm to tell whether the image is sensitive. Experimental results show that this method can detect the sensitive images under complex background effectively. The accurate rate can achieve 90.2%, the recall rate can achieve 86.3%, and the error rate can achieve 3.5%.

Key words: sensitive image, skin color detection, human body detection, Support Vector Machine(SVM), Histogram of Oriented Gradient(HOG)

中图分类号: