基于Canny边缘检测的自适应空域隐写术

Adaptive Spatial Steganography Based on Canny's Edge Detection

针对自适应空域隐写术设计的关键问题,该文结合Canny边缘检测和校验格编码(STC)提出一种不需要同步边信息的自适应空域隐写方法。首先,根据秘密消息长度、载体图像等因素确定Canny边缘检测算法中的参数取值,进而根据相应的参数取值使用Canny边缘检测算法来选择载体图像的边缘区域。然后,分别定义边缘区域像素和非边缘区域像素的嵌入失真;最后,在载体像素的多个最低有效位平面(LSB)使用STC嵌入秘密消息。实验结果表明:该隐写方法在4种嵌入率情况下抵抗常见通用隐写分析的性能优于3种已有的隐写方法,且在较小嵌入率情况下与空域通用小波相对失真方法(S-UNIWARD)相当。

Aiming at the essential problems of the design of adaptive spatial steganography, this paper proposes an adaptive spatial steganographic algorithm without synchronizing the side information, combining with Canny＇s edge detection algorithm and the Syndrome Trellis Code （STC）. Firstly, the parameters of Canny＇s algorithm are obtained on the basis of the factors, including the length of the secret message, the cover image, and so on; then Canny＇s algorithm is used to select the edge region of the cover image. Moreover, the embedding distortions of the edge and non-edge pixels are defined respectively. Finally, the STC is used to embed the secret message in multiple Least Significant Bit （LSB） planes of the pixels. The experimental results illustrate that, under the condition of four kinds of embedding rates, when resisting common universal steganalysis, the proposed method performs better than other three existing methods, and is comparable to the Spatial-UNiversal WAvelet Relative Distortion （S-UNIWARD） under the condition of small embedding rates.