.
RESEARCH PAPER
.
SCIENCE CHINA
Information Sciences
August 2014, Vol. 57 082314:1–082314:10
doi: 10.1007/s11432-014-5106-8
c
Science China Press and Springer-Verlag Berlin Heidelberg 2014 info.scichina.com link.springer.com
A channel selection rule for YASS
YU LiFang
1
, ZHAO Yao
1∗
, NI RongRong
1
& CAO Gang
2
1
Institute of Information Science, Beijing Jiaotong University, Beijing 100044, China;
2
School of Computer Science, Communication University of China, Beijing 100024, China
Received February 3, 2014; accepted May 10, 2014; published online June 23, 2014
Abstract In this paper, a channel selection rule for YASS (Yet-Another-Secure-Steganography) is proposed.
Secret message embedding imposes distortion to the cover image. The larger the distortion, the less secure the
steganographic algorithm. Our channel selection rule engages in minimizing this distortion brought in by YASS.
In our rule, the distortion caused by unit change on each quantized DCT (Discrete Cosine Transformation)
component is computed, and the components with smaller unit change distortion are selected with higher
priority. This channel sel ection rule reduces distortion to the medium spatial domain image and the final JPEG
image. Experimental results show that our improved YASS scheme outperforms original YASS scheme on the
aspects of both perception and statistics. This new channel selection rule can also be combined with other
enhancements in YASS framework to further boost the performance.
Keywords steganography, steganalysis, YASS, channel selection, DC components, embedding distortion
Citation Yu L F, Zhao Y, Ni R R, et al. A channel selection rule for YASS. Sci China Inf Sci, 2014, 57:
082314(10), doi: 10.1007/s11432-014-5106-8
1 Introduction
Steganography is the science and art of writing hidden messages so tha t no one apa rt from the intended
recipient knows the existence of the message. On the opposite side, steganalysis detects the existence of
the secretly embedded message, or even estimates the message le ngth, and finally possibly extracts the
hidden message.
Yet-Another- Se cure-Steganography (YASS) [1] hides secret messages in pseudo randomly se le cted do-
main and compresses the stego image into JPEG format. Hence, its embedding domain is not easy to be
decided and it is promising for high security performance. Compared to prosperously develope d passive
steganog raphy, YASS is a naive steganographic framework with no n-carefully considered details, e.g., the
selection of embedding domain, embedding channel, embedding algorithm, embedding and advertising
JPEG quality factor, and so on. Therefore, many efforts have been made to improve its performance.
Ref. [2] varies embedding JPEG quality factor a ccording to block varianc e to enhance security, and
utilizes an iterative hiding method to improve pure payloads. Refs. [3–5] further randomize the loca-
tions of embedding blocks, and they survive the spec ific steganalysis against YASS [6]. Ref. [7] use s an
improved version of QIM (Quantization Index Modulation) and Ref. [8] combines matrix encoding into
YASS framework and employs punctured RA (Repeat-Accumulate) codes. All of these effo rts improve
the performance of YASS and make steganography based o n YASS framework more and more attractive.
∗
Corresponding author (email: yzhao@bjtu.edu.cn)