一种用于非法图像检测的云计算Hadoop平台设计  

Illegal Image Detection Based on Cloud Computing Platform Hadoop

作　　者：史建政[1] 于东敏[2] 

机构地区：[1]廊坊职业技术学院,河北廊坊065000 [2]河北工业大学廊坊分校,河北廊坊065000

出　　处：《计算机测量与控制》2013年第6期1628-1630,1633,共4页Computer Measurement &Control

基　　金：河北省教育厅教学改革立项支持项目(103004);教育部高职委资助项目(jzw590111050)

摘　　要：互联网中存在着大量的非法图片严重影响了正常的网络生活,传统的非法图像识别主要是经过图像与样本库中的图像进行对比,工作量巨大,实时性能较差;提出基于云计算Hadoop平台的非法图像检测方法,将非法图像特征提取以后通过统计分析构造贝叶斯分类网络,将贝叶斯判断的方法MapReduce后部署在Hadoop云计算平台进行非法图像的分类;实验结果证明,这种方法的加速比效果明显增加,图像分类识别的准确率也大大地提高。The Internet in the presence of a large number of illegal images has seriously affected the normal life of the network. The tra ditional illegal image recognition is mainly through image and the sample of image contrast, heavy workload, poor realtime performance. Based on cloud computing platform Hadoop illegal image detection method, image feature extraction after illegally through statistical analysis to construct Bias classification network, will Bias judging method of MapReduce after deployment in Hadoop cloud computing platform for il legal image classification. The experimental results show that, this method speedup effect is significantly increased, image classification accu racy is greatly improved.

关 键 词：非法图像检测 云计算Hadoop平台 贝叶斯网络 

分 类 号：TP146[自动化与计算机技术—控制理论与控制工程]