A reversible image hiding scheme using novel linear prediction coding and histogram shifting Dae-Soo Kim 1 , Gil-Je Lee 2 , and Kee-Young Yoo 3 1 Department of Information Security, Kyungpook National Univ., Daegu, South Korea 1 Graduate School of Electrical Engineering and Computer Science, Kyungpook National Univ., Daegu, South Korea 3 School of Computer Science and Engineering, Kyungpook National Univ., Daegu, South Korea (Corresponding author : Kee-Young Yoo) Abstract— In 2009, Tsai et al. proposed reversible image hiding scheme using linear prediction coding and histogram shifting. Tsai et al.’s scheme improved the hiding capacity of Ni et al.’s scheme using the linear prediction coding and two histograms. In the linear prediction coding, however, the basic pixel is not used. If a value of basic pixel is the largest or the smallest in a block, only one histogram is generated and the hiding capacity is decreased. To solve the problems, this paper proposes the novel linear prediction coding with the inverse S-order and one histogram using two peak points. In experimental results, the hiding capacity of the proposed scheme is superior to Tsai et al.’s scheme. Keywords: Reversible data hiding, Steganography, Histogram shifting, Inverse S-order 1. Introduction Data hiding is a technique that embeds data into digital media to convey secret data by slightly altering the contents of the media, so that the embedded data is imperceptible [1], [2]. In image data hiding, during data embedding, distortion of image occurs since the pixel values in the cover image will be inevitably changed. The sensitive images, such as military images, medical images, or artwork preservation, are intolerable about the embedding distortion. For medical images, even slight changes are caused of the potential risk of the misdiagnosis. Nowadays, reversible data hiding research has become quite important. Reversible data hiding techniques are de- signed to solve the problem of sensitive images. After the embedded secret data is extracted, the image can be completely restored to its original state before embedding occurred. Several reversible data hiding schemes have been proposed [4], [5], [6], [7]. In 2006, Ni et al. proposed a very different reversible data hiding technique based on the histogram shifting technique [8]. Ni et al.’s scheme adjusts pixel values between peak point and zero point to conceal data and to achieve reversibility. However, the capacity is limited by the most frequent pixel values in the cover image. After proposed Ni et al.’s scheme, to improve the reversible data hiding based histogram shifting had researched [3], [9], [10], [11], [12], [13], [14], [15]. In 2009, Tai et al. proposed a pixel difference based reversible data hiding scheme [3]. Tsai et al. proposed a block-based reversible data hiding scheme using prediction coding [11]. However, Tsai et al.’s scheme had problems in prediction coding and dividing histogram into two sets. This problems cause a decrease of the hiding capacity. In this paper, the novel linear prediction coding with the inverse S-order and generating one histogram using two peak points are proposed to solve the problems of Tsai et al.’s scheme. The rest of this paper is organized as follows: In Section 2, the reversible data hiding schemes proposed by Tai et al. and Tsai et al. are introduced. In Section 3, the embedding, extraction and recovery procedures of the proposed method are presented. Experimental results are given in Section 4, and ending with conclusions in Section 5. 2. Related works 2.1 The pixel differences in an inverse S-order Tai et al. is proposed to a similar method in which the differences of two consecutive pixels are calculated. Data embedding is done by modifying the histogram of the absolute value of the differences using a proposed binary tree [3]. In Tai et al.’s scheme, scans the pixel value c i of cover image C in an inverse S-order as shown in Fig. 1. The pixel difference e i calculates between pixels c i−1 and c i as follows e i =  c i , if i=0 |c i−1 - c i |, otherwise (1) 2.2 The histogram shifting Ni et al. proposed a histogram-based reversible data hiding scheme [8]. In their scheme, all pixel values in the cover image are calculated to generate the image histogram for secret data embedding. The peak point and the zero