Chapter 5 Creation of Variable-Sized Mosaic Images for Information Hiding .56
5.3 Proposed Method for Data Hiding in Variable-Sized Mosaic Images
5.3.1 Concept of Proposed Method
The main concept of the proposed method for data hiding in a variable-sized mosaic image is to utilize the variability of the size of the tile image. We can hide data bits into the tile images by taking a bit sequence of a given secret message as an input array A to produce an image by applying the proposed image creation process described previously. In the data extraction part, detecting the tile size with no borders from the resulting image is too hard, so we proposed a method to solve this problem.
We must modify slightly the specific pixel of each tile image. If the tile size is large, we set the RGB values of the specific pixel to the even values nearest to the original ones, respectively; and if the tile size is small, we set them to the odd values nearest to the original ones, respectively. The details will be stated clearly in Sections 5.3.2 and 5.3.3.
5.3.2 Data Hiding Process
As mentioned previously, the main idea behind the proposed data hiding process is to utilize the tile size according to an array with hidden bits. Simply speaking, we get the appropriate size for the tile image to be drawn according to a hidden bit by checking the matching table shown in Table 5.1. In order to avoid producing overlapping areas, we enforce the tile size to be adjusted in above-mentioned manner.
Then, we modify slightly a specific pixel in the tile image to mark the tile size by Algorithm 5.3. Finally, we compose all the tile images together to produce a stego-variable-sized mosaic image. The detail of the data hiding process is described
as an algorithm as follows.
Algorithm 5.2: data hiding in a variable-sized mosaic image.
Input: an image I, a secret message Mesi, and a secret key K.
Output: a stego-variable-sized mosaic image for I’. Steps:
Step 1. Create an empty array A.
Step 2. Convert the secret message size into a bit sequence with a size of 10 and add it to A in order.
Step 3. Take Mesi and a secret key K to derive a disordered data DDatai by Algorithm 3.3 in Section 3.3.2.
Step 4. Add DDatai to A in order.
Step 5. Take A as an input array to run the image creation process proposed in Section 5.2.2.
Step 6. Modify slightly a specific pixel of each tile image by Algorithm 5.3.
Algorithm 5.3: modify the RGB values of a specific pixel.
Input: a tile image T and a secret key K.
Output: a modified tile image T’. Steps:
Step 1. Use K as an input to run a random number generator and get a random integer RI ranging from 0 to 255.
Step 2. Get the coordinates of a specific pixel P in the following way:
2.1. Take the x coordinate of P to be the value of RI divided by 16.
2.2. Take the y coordinate of P to be the remainder of RI divided by 16.
Step 3. Modify the RGB values of P in the following way:
3.1. if the tile size is large, set the RGB values of P to the even values nearest to the original ones, respectively;
3.2. otherwise, set the RGB values of P to the odd values nearest to the original ones, respectively.
5.3.3 Data Extraction Process
The main concept of the data extraction process is to figure out the size of each tile. It is the most important issue that not every tile hides a bit. Therefore, we must judge whether a bit is hidden or not into a tile. If we met a space which cannot fit a large tile, then this tile contain no hidden bit. The data extraction process is described as an algorithm as follows.
Algorithm 5.4: data extraction process.
Input: a stego-variable-sized mosaic image S and a secret key K.
Output: a bit sequence of a secret message Mesi. Steps:
Step 1. Get the secret message size from the first ten tile images.
1.1. Use Algorithm 5.5 to process the first ten tile images and store the extracted bits to an array A1 in order.
1.2. Convert the array A1 into a decimal value which is just the secret message size MSize.
Step 2. Use Algorithm 5.5 to process other tile images and add the extracted bits to an array A2 in order.
Step 3. Take A2 and the secret key K to derive Mesi by Algorithm 3.8 in Section 3.3.5.
Algorithm 5.5 extracting a hidden bit from a tile image.
Input: a tile image T and a secret key K.
Output: a hidden bit B.
Steps:
Step 1. If there is not enough space to insert a large tile, go to Step 2, otherwise go to Steps 3.through 6.
Step 2. There is no hidden bit can be extracted from this tile image.
Step 3. Use K as an input to run a random number generator and get a random integer RI ranging from 0 to 255.
Step 4. Use Step 2 of Algorithm 5.3 to get the coordinates of the specific pixel P of T.
Step 5. Judge the RGB values of P are even or odd. If the values are even, the tile size is large; otherwise, the tile size is small.
Step 6. Take the tile size to check the matching table shown in Table 5.1 to get a hidden bit B.
5.3.4 Experimental Results and Discussions
Figure 5.6, Figure 5.7, and Figure 5.8 show some experimental results of data hiding in variable-sized mosaic images. Figure 5.6(a) is a stego-variable-sized mosaic image with the secret message, “巴黎鐵塔.” Figure 5.6(b) is the secret message extracted from Figure 5.6 (a) with a correct key. Figure 5.6(c) is the secret message extracted from Figure 5.6 (a) with a wrong key. Figure 5.7(a) is a stego-variable-sized mosaic image with the secret message, “狒狒 baboon.” Figure 5.7(b) is the secret message extracted from Figure 5.7(a) with a correct key. Figure 5.7(c) is the secret message extracted from Figure 5.7(a) with a wrong key. Figure 5.8(a) is a stego-variable-sized mosaic image with the secret message, “白雪皚皚.” Figure 5.8(b) is the secret message extracted from Figure 5.8(a) with a correct key. Figure 5.8(c) is the secret message extracted from Figure 5.8(a) with a wrong key.
(a)
(b) (c)
Figure 5.6 Experimental results of data hiding in variable-sized mosaic image. (a) A
stego-variable-sized mosaic image. (b) The secret message extracted from (a) with a correct key. (c) The secret message extracted from (a) with a wrong key.
(a)
(b) (c)
Figure 5.7 Experimental results of data hiding in variable-sized mosaic image (a) A
stego-variable-sized mosaic image. (b) The secret message extracted from (a) with a correct key. (c) The secret message extracted from (a) with a wrong key.
(a)
(b) (c)
Figure 5.8 Experimental results of data hiding in variable-sized mosaic image. (a) A
stego-variable-sized mosaic image. (b) The secret message extracted from (a) with a correct key. (c) The secret message extracted from (a) with a wrong key.
Chapter 6
Conclusions and Suggestions for Future Works
6.1 Conclusions
In this study, we have proposed three methods for creation of art images of new types and data hiding in them. We can apply these methods for various information hiding applications, such as covert communication, copyright protection, etc.
Different from traditional data hiding techniques, we hide data in the individual features of the created art images.
The three different types of art images created in this study are irregular-hexagonal-tiled image, tile-overlapping mosaic image, and variable-sized mosaic image.
For irregular-hexagonal-tiled images, we use two features for image creation and data hiding. One is the variation of two specific vertices of a hexagonal tile, and the other is the RGB values of the center of a tile. We combine a secret message and a watermark to form a bit sequence before performing the data hiding process. By modifying the locations of the two specific vertices and the RGB values of the center of every hexagonal tile, we can achieve the data hiding work.
In tile-overlapping mosaic images, we use one feature for image creation and data hiding, that is, the overlapping degree of every pair of adjacent tile images. Due to some special cases of tile arrangements, holes might be created during the image creation process. We solve this problem by restricting the tile location in a specific
range between the current tile and the preceding one of the left and upper side. We can achieve the data hiding work by an overlapping scheme which utilizes the overlapping degrees of adjacent tiles.
For variable-sized mosaic images, we use only one feature for image creation and data hiding, that is, the size of the tile image. We also proposed a solution to avoid causing overlapping areas during the image creation process. The solution is based on checking the sizes of three neighboring tiles to see whether there is enough space to insert the large tile. By adjusting the tile size according to a bit sequence of given message data, we can achieve the purpose of data hiding.
6.2 Suggestions for Future Works
In this study, we have proposed some methods for creation of irregular-hexagonal-tiled images, tile-overlapping mosaic images, and variable-sized mosaic images, as well as data hiding techniques for these three types of art images.
The two applications, secret communication and copyright protection, may be carried out by utilizing the proposed data hiding techniques. However, there are still some interesting topics which are worth for further study. They are listed as follows.
1. Producing mosaic images with other shapes based on computer graphics.
2. Combining two proposed data hiding methods and applying them to a mosaic image for a larger bit hiding capacity.
3. Using more tile features for embedding more data into mosaic images.
4. Detecting the tile borders to speed up data extraction by applying a more complicated image analysis process for a mosaic image composed of many tile images.
5. Creating other types of art images and selecting appropriate image features to achieve corresponding data hiding works.
6. Keeping the image quality good after embedding data in it.
References
[1] Y. W. Guo, J. H. Yu, X. D. Xu, and Q. S. Peng, “Example based painting generation,” Journal of Zhejiang University: Science, Vol 7, No 7, pp.
1152-1159, July 2006.
[2] G. C. Chen, “Automatic Generation of Pencil Sketching with the Effects of Paper Texture,” M. S. Thesis, Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan, Republic of China, June 2006.
[3] B. Baxter, V. Scheib, M. C. Lin, and D. Manocha, “DAB: Interactive haptic painting with 3D virtual brushes,” Proceedings of SIGGRAPH 01, pp. 461-468, Auguest 2001.
[4] Von Laerhoven, Tom and Van Reeth, Frank, “Real-time simulation of watery paint,” Computer Animation and Virtual Worlds, Vol 16, No 3-4, pp. 429-439, July 2005.
[5] A. Hertzmann, “Painterly Rendering with Curved Brush Strokes of Multiple Sizes,” Proceedings of SIGGRAPH 98, pp. 453-460, Orlando, Florida, July 1998.
[6] C. J. Curtis, S. E. Anderson, J. E. Seims, K. W. Fleischer, and D. H. Salesin,
“Computer-generated watercolor,” Proceedings of SIGGRAPH 97, pp. 421-430, August 1997.
[7] U. N. Chen, “Non-Photorealistic 3D Rendering in Chinese Painting Style,” M. S.
Thesis, Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan, Taiwan, Republic of China, July 2003.
[8] K Kojima, S Takahashi, T Nishita, “Creating Quadrilateral Mosaics from Image
Topographic Features,” Proceedings of the 3rd symposium on Applied perception in graphics and visualization, 2006.
[9] Paul E. Haeberli, “Paint by Numbers: Abstract Image Representations,”
Proceedings of SIGGRAPH 90, pp. 207-214, August 1990.
[10] Alejo Hausner, “Simulating Decorative Mosaics,” Proceedings of SIGGRAPH 01, pp. 573-580, New York, USA, 2001.
[11] Hoff, K., Keyser, J., Lin, M., Manocha, D. and Culver, T. “Fast Computation of Generalized Voronoi Diagrams Using Graphics Hardware,” Proceedings of SIGGRAPH 99, pp. 277-286, August 1999.
[12] S. P. Lloyd, “Least Square Quantization in PCM,” IEEE Transactions on Information Theory, Vol. IT-28, No. 2, pp. 129-137, March 1982.
[13] D. C. Wu and W. H. Tsai, “A Steganographic Method for Images by Pixel-Value Differencing,” Pattern Recognition Letters, Vol. 24, No. 9-10, pp. 1623-1636, 2003.
[14] Y. C. Chiu and W. H. Tsai, “Copyright protection by watermarking for Color Images against Rotation and Scaling Attacks Using Peak Detection and Synchronization in Discrete Fourier Transform Domain,” Proceedings of Third Workshop on Digital Archives Technologies, pp. 207-213, Taipei, Taiwan, August 5-6, 2004.
[15] Z. Ni, Y. Q. Shi, N. Ansari, and W. Su, “Reversible data hiding,” Proceedings.
IEEE Int. On Cirsuits and Systems, Vol. 2, pp. 912-915, Bangkok, Thailand, May 25-28, 2003.
[16] Y. J. Cheng, C. Y. Yin, D. C. Wu, and W. H. Tsai, “Copyright protection and authentication of image information in digital libraries,” Proceedings of 2001 Conference on Trends and Perspectives in Digital libraries and Digital Museums, Vol 3, pp. 1-11, Hsinchu, Taiwan, Republic of China.
[17] W. L. Lin, “Data hiding in image mosaics,” M. S. Thesis, Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan, Republic of China, June 2004.
[18] S. C. Hung, “Data hiding in art images,” M. S. Thesis, Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan, Republic of China, June 2005.
[19] C. Y. Hsu and W. H. Tsai, “Creation of a new type of image - circular dotted image - for data hiding by a dot overlapping scheme,” Proceedings of Conference on Computer Vision, Graphics and Image Processing, Taoyuan, Taiwan, Republic of China, 2006.
[20] C. Y. Hsu, “A study on art image generation and information hiding,” M. S.
Thesis, Department of Computer Science and Engineering, National Chiao Tung University, Hsinchu, Taiwan, Republic of China, June 2006.