A New Method for Data Hiding Domain Classification
Jing Liu, Tang Guangming · 2009
Data hiding domain identification is very important for further steganalysis. A statistic feature-based method is advocated for identifying the hiding domain in this research. Different features in different domains that can reflect well statistical changes due to data hidden are extracted. In order to obtain better classification results, a process of adding noise is introduced to extract the feature that can classify spatial and DCT domain hiding. Based on the "one-against-one" SVM classifier, the experiment is implemented and the proposed method has performed satisfied classification results.