Which gray level should be given the smallest cost for adaptive steganography?

Currently, the most successful approach to steganography in digital image is distortion-minimization framework, which reduces the steganographers' work to the design of distortion function with the aid of practical coding schemes. Previous distortion functions for spatial images are all position dependent, in which cost is determined by the relationships between neighboring pixels. Noticing that Gamma encoding is usually involved in image preprocessing in many cameras or image processing software, which causes some pixels to change greatly, we believe these pixels sensitive to Gamma encoding are more suitable for modification, because they are hard to model due to their large variations. Inspired by this idea, we proposed a position independent scheme, where the cost is only linked to the gray level. The effectiveness of our work is verified by extensive experimental results, which reveal an interesting relationship between steganographic costs and gray levels. The speed test shows that the speed of proposed scheme is very high thus suitable to be used in the real-time applications.