几种模式识别方法在源相机辨识中的分析比较  

Analysis and comparison of several pattern recognition methods in source camera identification

作　　者：洪艳[1,2] 

机构地区：[1]合肥工业大学信息与网络中心,合肥230009 [2]合肥工业大学计算机与信息学院合肥230009

出　　处：《电子测量与仪器学报》2012年第4期367-371,共5页Journal of Electronic Measurement and Instrumentation

摘　　要：为了重新衡量模式识别方法在整个源相机辨识过程中所起的作用,对在源相机辨识中使用比较普遍的几种模式识别的方法,包括支持向量机、BP神经网络以及Simple Stump Adaboost,进行了分析比较。通过对提取到的300张照片的特征数据分别采用了分析实验,根据得到的实验结果,在通用性、效率、推广能力和分类效果方面对各种算法进行分析与比较,最后得出每种方法的优缺点和适用性,为源相机辨识的实际应用提供有利的信息。In order to re-evaluate the method to analyze the effect of entire source camera identification process, the analysis and comparison for several popular pattern recognition methods used in source camera identification, including SVM, BP neural network and Simple Stump Adaboost, is carried out. Through the analysis experiment on the extracted feature data of 300 photos, and according to the result of analysis and comparison to versatility, efficiency, promotion abil- ity and classification ability of each algorithms, the advantages and disadvantages of each method is finally gotten, which offers the favorable information for the practical application of the source camera identification.

关 键 词：源相机辨识 特征提取 支持向量机 神经网络 集群算法 模式识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]