Multi-layer Progressive Secret Image Sharing Wen-Pinn Fang Department of Computer Science and Information Engineering Yuanpei University No.306, Yuanpei St., HsinChu Taiwan 30015, R.O.C wpfang@mail.ypu.edu.tw Abstract: - This paper proposes a special type secret image sharing method. Based on the theories of secret image sharing and frequency domain transform, a multi-threshold secret image sharing is achieved. The method contains more properties compared with conventional secret image sharing which just has only one threshold. The properties of the proposed method include fault tolerance, small size of shares, secure, multi-threshold and progressive transmission. The method not only has the advantages of conventional secret image sharing, but also is more flexible. Users can set several thresholds in the method. When the number of shares is less than the smallest threshold, no secret will be revealed. While the numbers of shares collected are more than smallest threshold, a low quality secret image will be revealed. The quality of recovered secret image will be better after collecting more shares. When the amount of shares is over the maximum threshold, the recovered secret image is the same as the original secret image. Key-Words: - Secret sharing, Progressive, Share, Fault-tolerance, Threshold, Transmission 1 Introduction In an (r, n) image sharing system [1-3], n shares {L 1 , L 2 ,..., L n } are created for a given image, e.g., Lena. The image can be revealed when r shares are received, while less than r shares reveal nothing about the image. With only sharing, nobody (even the company organizer) can view the image without attending a public meeting. Therefore, sharing is a valuable safety process especially in a company where no employee/investor alone should be trusted. Significantly, the original image can be discarded after sharing; moreover, each of the r shares is 1/r of the size of the given image. Therefore, the sharing process does not waste storage space. Dissimilar to the traditional secret image sharing, the method proposed in this paper can control the amount of information released by means of the amount of shares collected. With this feature, a majority rule will be established. For example, if a dealer has some shares but his share is a noisy image, we can say that he has no right to get the information. If there are two dealers with individual share images, the one with the more informative share has more power. The rest of this paper is organized as follows: the background knowledge is show in Section 2; the method is proposed in Section 3; Quality Control is discussed in Section 4, Experimental results are shown in Section 5. Finally, the discussion is represented in Section 6. 2 Background knowledge Before describing the method, it is necessary to explain some basic knowledge. The basic knowledge includes discrete cosine transform, zig-zag scan and the kernel of secret image sharing. The detail is shown as below: 2.1 Discrete cosine transformation (DCT) Discrete cosine transform is one of the most frequently used transformation for image compression, Equation(1) is a 2-D DCT equation for 8×8 non-overlapping block. ) , ( 16 ) 1 2 ( cos 16 ) 1 2 ( cos 4 ) ( ) ( ) , ( 7 0 7 0 j i f v j u i v c u c v u F i j                     (1)         . 0 , 1 , 0 , 2 1 ) ( e if e if e c Here, F(u,v) and f(i,j) present a DCT coefficient at the coordinate (u,v) and a pixel value at the coordinate (i,j), respectively. F(0,0) is called the direct current (DC) component, which corresponds to Proceedings of the 7th WSEAS Int. Conf. on Signal Processing, Computational Geometry & Artificial Vision, Athens, Greece, August 24-26, 2007 112