An optimized medical images watermarking technique utilizing a constant ratio of frequency domain

Abstract

A medical image is a visual representation of the shape or function of internal body organs or tissues used in clinical analysis or medical diagnosis. Watermarking techniques are used to protect copyright and provide evidence of any attempt to unlawfully manipulate the content of medical images. The watermarking technique was used to mask the data within the medical image without affecting the original visual quality. The challenge is producing an optimal watermarked image for different types of medical images where each with unique characteristics. Currently, there is no watermarking technique that can find the optimal action to use without affecting the visual quality and can resist image attacks. This research aims to investigate the characteristics of the frequency watermarking technique that affect the process of producing an optimal watermarked medical image. In the proposed optimal watermarking technique, watermark data will embed in a constant ratio of the original medical images frequency domain to increase the resistibility of the watermarked medical image to image attacks. The first step in developing an optimal watermarking technique in this research is to construct a theoretical framework and conceptual design model to understand the underlying characteristics of the watermarking technique process. Based on the proposed theoretical framework and conceptual design model, this research then investigates the role of the optimized frequency technique to achieve optimal performance for the watermarking technique. This research conducted an implementation experiment and simulation for the optimized frequency watermarking algorithm. Also, it analysed the effect of using the optimized frequency domain on the performance of the watermarking technique. Finally, the results of the optimized frequency watermarking algorithm are compared to the results of published watermarking algorithms. The proposed watermarking technique used optimization algorithm with frequency technique and Schur decomposition to produce watermarked images able to trade-off between the performance evaluation terms. The imperceptibility performance is high and has the robustness of the JPEG compression, salt and pepper noise, Gaussian noise, and rotation attack image attacks. Where the average PSNR values is 75.44 dB and NCC values for rotation attack is 0.91 and PEG compression is 0.99.