Cover Image Selection Technique for Secured LSB-based Image Steganography

Abstract

Various researches in image steganography focusing on spatial domain specifically in the least Significant Bit (LSB) embedding method had been conducted to improve the embedding capacity while maintaining high imperceptibility. However, these improvements were countered by various statistical attacks. This paper presents an LSB-based image steganography technique that uses YCbCr color space for the embedding process. Also, a new cover selection method to strengthen the proposed embedding algorithm was introduced in this paper. The cover selection mechanism used the skewness and kurtosis of the candidate cover images as factors to determine if a candidate cover image will yield to a high probability of detection or not. The distortions analysis affirms that the stego-images produced by the embedding method obtained acceptable PSNRs and SSIMs, thus, proving that the stego-images are resistant to Human Visual System. Also, the embedding method presented in this paper produced stego-images that were given a low probability of detection of various statistical analysis attacks. Furthermore, the high correlation coefficient between the candidate cover images' skewness and kurtosis and the probability of detection obtained by the stego-images affirms that these properties of the candidate cover images can help in determining the most suitable candidate cover images to be used.