计算机技术与发展2016，Vol.26Issue(10)：1-5,5.DOI:10.3969/j.issn.1673-629X.2016.10.001
针对图像来源鉴别中支持向量机的研究
Research on Support Vector Machines for Image Source Identification
摘要
Abstract
With the popularity of digital images,blind image forensics has become one of the hotspots nowadays. The main research con-tent of blind image forensics is how to identify the image source. As the most critical stage of image source identification,the SVM classi-fication model for identification directly affects the final identification rate. Because the different kernel function and kernel parameters has distinct effect on the performance of the classification model,the various kernel functions are analyzed and compared,then the Gaussian radial basis function with better subdivision is selected as the kernel function. In view of the kernel parameter selection,the various kernel parameter optimization algorithms are analyzed,and the effectiveness of each algorithm and the effect of the final classification model by experiments are verified. The results show that choosing Gaussian radial basis function as the kernel function,using the kernel parameters selected by particle swarm algorithm to construct the classification model will achieve the best image source identification rate.关键词
图像盲取证/支持向量机分类模型/核函数/核参数/图像来源鉴别率Key words
blind image forensics/SVM classification model/kernel function/kernel parameter/image source identification rate分类
信息技术与安全科学引用本文复制引用
黄曜,许华虎,欧阳杰臣,高珏..针对图像来源鉴别中支持向量机的研究[J].计算机技术与发展,2016,26(10):1-5,5.基金项目
上海张江国家自主创新示范区专项发展资金重点项目(一期)(201411-ZB-B204-012) （一期）
