作者: J. Hemalatha , M. K. Kavitha Devi , S. Geetha
DOI: 10.1007/S10586-017-1500-5
关键词:
摘要: The major challenge of feature based blind steganalysers lies in designing effective image features which give true evidence the stego noise rather than natural present images. Hence they report low detection accuracy real time implementation spite employing 100s process. In this paper, we coin a new paradigm for detecting steganography by examining task as three-steps process with following repercussions: (a) curvelet transform denoising pre-processing step that produces better residuals suppressing residual general before extraction, (b) extracting various steganalytic features, both spatial domain well and (c) implementing system on an efficient classifier, multi-surface proximal support vector machine ensemble oblique random rotation forest, provides rate superior to other existing classifiers. Extensive experimentation huge database clean steganogram images produced from seven steganographic schemes varying embedding rates, using five steganalysers, shows proposed improves substantially proves be high performance strategy even at rates. This model can employed preprocessing component any steganalyser obtained.