Large scale image steganalysis based on MapReduce

Abstract

Steganalysis is the opposite art to steganography, whose goal is to detect whether or not the seemly innocent objects like image hiding message. Image steganalysis is important research issue of information security field. With the development of steganography technology, steganalysis becomes more and more difficult. Regarding the problem of improving the performance of image steganalysis, many research work have been done. Based on current research, large scale training set will be the feasible means to improve the steganalysis performance. Classic classifier is out of work to deal with large scale images steganalysis. In this paper, a parallel Support Vector Machines based on MapReduce is used to build the steganalysis classifier according to large scale training samples. The efficiency of the proposed method is illustrated with an experiment analysis.