Introduction to the special section on image processing in security applications (VSI-ipsa)

Eduardo Cabal-Yepez, Luis Manuel Ledesma-Carrillo, Guillermo Botella, Irene Amerini

Abstract: The integrity and authenticity of electronic patient records (EPRs) are essential concerns in the healthcare industry. To address these concerns, this paper introduces a novel, robust watermarking scheme using integer wavelet transform-singular value decomposition (IWT-SVD). The proposed scheme is divided into logo embedding and EPR text embedding, aiming to mitigate false-positive issues and maintain EPR integrity and authenticity. In the first part, medical image and watermark logo are encrypted using DMA (Diffused Mandelbrot set-Arnold map) to enhance security. Further, the encrypted medical image is segmented into four sub-bands (CA, CH, CV, CD) by one-level IWT transform. Then SVD is applied on sub-band CD to extract the principal component (PC), where the encrypted logo is embedded due to its higher magnitudes, thereby mitigating false-positive errors. Embedding the encrypted logo into the hybrid IWT-SVD domain enhances security and robustness. Moreover, for embedding, multiple optimal embedding factors (MOEFs) are determined through Non-dominated Sorting Genetic Algorithm II (NSGA-II) to balance imperceptibility and robustness. In the second part, EPR text is embedded into watermarked logo using SVD to verify authenticity. The performance of the proposed scheme is validated on standard medical image datasets against various attacks in terms of SSIM, PSNR, BER, and NC. The proposed scheme improves PSNR values by 48.44% and 19.66% for grayscale and color images, respectively. Similarly, NC values are also improved by 10%–12% for grayscale and color images. Therefore, the proposed scheme enhances security, imperceptibility, and robustness while reducing complexity compared to the state-of-the-art schemes. Moreover, it is also false-positive-free and maintain integrity simultaneously.