Blind MV-based video steganalysis based on joint inter-frame and intra-frame statistics

• DOI: 10.1007/s11042-020-10001-9
• Publication Date: 2020-11-08T14:02:33Z
• Authors: Negin Ghamsarian; Klaus Schoeffmann; Morteza Khademi
• Keywords: Steganalysis; Overfitting; Codec; Motion vector; Feature vector

Abstract Despite all its irrefutable benefits, the development of steganography methods has sparked ever-increasing concerns over abuse in recent decades. To prevent inimical usage steganography, steganalysis approaches have been introduced. Since motion vector manipulation leads to random and indirect changes statistics videos, MV-based video center attention years. In this paper, we propose a 54-dimentional feature set exploiting spatio-temporal features vectors blindly detect stego videos. The idea behind proposed originates from two facts. First, there are strong dependencies among neighboring MVs due utilizing rate-distortion optimization techniques belonging same rigid object or static background. Accordingly, MV can leave important clues on differences between each blocks. Second, majority original videos locally optimal after decoding concerning Lagrangian multiplier, notwithstanding information loss during compression. Motion alteration embedding affect these that be utilized for steganalysis. Experimental results shown our features’ performance far exceeds state-of-the-art methods. This outstanding lies utilization complementary affected by as well dimensionality reduction applied overfitting. Moreover, unlike other existing methods, adjusted various settings codec standards such sub-pixel estimation variable-block-size estimation.