Development of effective information-hiding method for embedded systems

Abstract

This paper reports on the JPEG compression tolerance of novel proposed digital watermarking algorithm that uses the morphological wavelet transform based on max-plus algebra. For objective evaluation, we investigated the image quality and tolerance against graphical data compression attack of an image during the digital-watermarking contest presented by the Committee for Information Hiding and its Criteria (IHC). During benchmarks of this contest, graphical data compression was done to reduce the file size of original image (45.5 MB) to 1/10 or less and the error rate of embedding data was reduced to 0%. As a result, the embedded data could be completely extracted. The peak signal-to-noise ratio (PSNR) was 54.08-54.66 dB. Afterwards, additional graphical data compression was done to reduce the file size to 1/20 or less of the original picture image and the error rate was reduced to 0%, and the PSNR was 52.75-54.35 dB. By compressing the image by, about 99%, the smallest data size attained was 0.49 MB and the PSNR was about 52 dB and maintained very high image quality with 0% of error rate. From these results, the processing time of the proposed algorithm was confirmed to be about 2.2 to 148.6 times faster than that of conventional algorithms on embedded systems.

Morphological Methods in Image Processing for Embedded Processors

“…The value of α is chosen by considering the trade-off between image quality and extraction accuracy. On the basis of experiments with several values of α, we have found that 5 is a practical value for α [2], [21].…”

“…This is because, in the case of embedding information data in digital contents for mobile applications, real-time processing speed is more important than data security and image quality. With this background, we report max-plus algebra-based morphological wavelet transform (MMT) watermarking as a novel data-embedding algorithm for highly parallel processing with a mobile embedded processor [2], [18], [21]. Figure 18 shows the watermarking flow using the MMT for the case of a 2 × 2 pixel sampling window.…”

Development of Compression Tolerable and Highly Implementable Watermarking Method for Mobile Devices

“…Consequently, the algorithm for the MMT is assumed to be suitable for embedded low-power processors. Utilizing these features, MMT watermarking has been proposed for mobile embedded devices in our papers [9], [10]. Figure 1 shows the watermarking flow using the MMT for a 2 × 2 pixel sampling window…”