Research on Information Hiding Technology Based on Color Digital Still Images(5)

I used Borland C++ to write a simple application "e show" that implements the above algorithm. Figure 4.3 is an original Lena image as a signature signal. On the left is the image without embedded hidden information. The right side is in the left image. Embedding the effect of Figure 4.3, you can see that it has almost no change with respect to the left figure.

Figure 4.4 Image Changes Before and After Adding a Signature Signal

in conclusion:

Since the original 24-bit BMP image file hides information, the byte value of its data part changes to 1 at most, and the pixel represented by this byte only changes by a maximum of 1/256. Therefore, the BMP image with hidden information and the BMP with unhidden information are hidden. Image, with the naked eye can not see the difference;

The advantage of embedding information directly into pixel RGB values ​​is that the capacity of the embedded information is proportional to the size of the selected masked image;

Using this method, a 24-bit BMP image file with a size of 32 k can hide information about 32 k/8 = 4 k (ignore 54 bytes where the file header cannot hide data). This method has high information hiding. rate.

5 Research Dynamics and Development Status of Information Hiding Technology

At the 1994 IEEE International Conference on Image Processing (ICIP`94), RG Schyndel and others first explicitly proposed the concept of “digital watermarking”[4], which triggered a culmination of modern information hiding technology research. Just two years later, on the ICIP `96, there have been research topics focusing on watermarking, copyright protection, and access control of multimedia services in the information hiding field. . In the same year, the first International Workshop on Information Hiding (First International Workshop on Information Hiding) was held in Cambridge, England, covering topics such as data hiding, secure communication, and cryptography. In the United States, many famous universities and research institutes of large companies, such as the MIT Media Lab, the University of Minnesota, Princeton University, the University of Southern California, etc., as well as NEC Corporation, IBM Corporation, etc., have all been working on information hiding technology. Research in the field has achieved a lot of research results. At the same time, a large number of digital watermarking software applications have emerged, such as HIGHWATER FBI, Digimarc Corporation, Fraunhofer`s SYSCOP and so on. At present, the watermarking technology applied to digital images has been considered to be written in the JPEG 2000 standard [5], which will surely promote the development of information hiding technology.

The existing digital watermarking algorithms mainly include the following types:

The least significant bit algorithm (LSB). LFTurner and RG van Schyndel et al. have used this algorithm to hide specific markers in digital audio and digital images. This is a typical spatial domain data hiding algorithm. Because the algorithm hides information by adjusting the lowest bits of the original data, it makes it difficult for the average user to visually and audibly detect hidden information. Nowadays, some simple information disguise software on the Internet mostly uses LSB technology and palette adjustment and other related technologies to hide information in 24-bit images or 256-color images, such as StegoDos, White Noise Sterm, and STools. Although it has a large amount of information hiding, as a digital watermarking algorithm, the algorithm is limited due to its basic principle, and the hidden digital watermarking information is extremely fragile and cannot withstand some lossless and lossy signal processing.

Document structure fine-tuning method. Brassil et al. proposed three techniques for hiding specific binary information in a general-purpose document (Postscript). Digital information is encoded by slightly adjusting the following structure in the document, including: vertical line spacing, horizontal kerning, and text character adjustment. (eg fonts). The algorithm can resist some standard document operations, such as photocopying and scan copying, but the technique is also easily destroyed by experienced attackers and is only suitable for document image classes.

Tanaka et al. studied the embedded digital watermark design similarly to the quantization noise, because the quantization noise in the digital signal is usually difficult to find by the observer. They embed digital watermarks by controlling the choice of the quantization level of the predictive quantizer, and the close-in data introduced therein approximates a quantization noise for the image; for a dithered image, the watermark is designed as a dither matrix.

Patchwork method, which is a statistic-based data watermark embedding scheme proposed by Bander et al. Patchwork arbitrarily selects N for the image points, while increasing the brightness, and correspondingly lowering the brightness value of the other point. Watermark embedding is accomplished through this adjustment process. This algorithm is not detectable and is resistant to lossy compression coding (JPEG) and some malicious attacks.

The texture block map coding method, which is also proposed by Bander et al., hides the digital information in any texture part of the digital image. This algorithm is resistant to filtering, compression and twisting operations, but it is only suitable for images with a large number of arbitrary texture areas, but it is not yet fully automated.

Frequency domain digital watermarking method. This is a way to hide data using techniques like spread spectrum communications. This technique hides hidden information in certain frequency domain modulations and hides it in the most perceptible spectral components of digital media. This technique can combat lossy compression coding and other data processing processes with signal distortion.

As a technical system, digital watermarking is still not perfect, and the angle of involvement of each researcher is different. Therefore, research methods and design strategies are also different, but all are designed around the basic characteristics of digital watermarking. At the same time, with the development and application of this technology, advanced technologies and algorithms in some other fields will also be introduced to complete and enrich the data watermarking technology. For example, wavelet and fractal theory in digital image processing, various compression algorithms in image coding, audio and video coding techniques, and the like.

Digital watermarking technology is a multidisciplinary emerging technology that spans signal processing, digital communications, cryptography, and computer networks. It has a huge potential application market and its research has important academic and economic value.

The information hiding technology represented by digital watermarking has also aroused strong interest from researchers in China. In order to promote the research and application of domestic information hiding technology, the three academicians in the field of information security in China jointly initiated the seminar on information hiding in China and organized the first session in December 1999. At the same time, the National 863 Program Intelligent Computer Expert Group held the "Digital Watermarking Technology Symposium" in January 2000. The seminar was hosted by the State Key Laboratory of Pattern Recognition of the Institute of Automation of the Chinese Academy of Sciences and the Information Security Center of Beijing University of Posts and Telecommunications. Participants conducted a comprehensive and in-depth discussion on the development and trends of digital watermarking technology.


Author/Wei Weimin School of Computer Science, Wuhan University