A wavelet based medical image watermarking scheme for secure transmission in telemedicine applications

Watermarking is one of the techniques used to protect multimedia data, and images in particular, from malicious attacks by inserting a signature into these images. However, traditional watermarking schemes encounter limitations for sensitive images such as medical images. Indeed, the inserted mark often alters the image in an irreversible way and can hide subtle details. In order to overcome these limitations, we propose in this paper a watermarking approach for patient identification and watermark integrity verification. In this approach, the watermark consists of two parts, the first part contains the patient's information fingerprint, and the second part contains the patient's encrypted photography. For the integration process, a discrete wavelet transform is applied to divide the medical image into four sub-bands. The watermark bits are then integrated by modulating the obtained mid-frequency coefficients. This approach's capacity allows the integration of the patient's photography and fingerprint, but will also be sufficient for a possible error-correcting code addition. The coefficient modulation performed by the integration process allowed hiding the watermark in an imperceptible way. This allowed us to generate a watermarked image structurally similar to the original. Performed experiments on the test images allowed to evaluate the performance of this method and to assess its robustness against JPEG compression and noise addition. Nevertheless, these values are sometimes a little far from the ideal value of 1 for normalized correlation.