Watermarking skin lesion digital color images based on channels
Simona Moldovanu, Diana Ştefănescu, Felicia Anisoara Damian Michis · 2020
This paper presents a way to secure through copyright medical images. The novelty of this study consists in splitting in color layers of a mole digital image and of a melanoma digital image, where the common mole is considered as the host image and the melanoma image as watermarked image. This work attempts an embedding of the information with DWT (discrete wavelet transformation) of the Haar wavelets on each color layer, decomposing them on one level, selecting the coefficient of level with low frequency to be the watermarking key and using IDWT ( inverse discrete wavelet transform) Haar wavelets for obtaining the invisible watermarked images. The efficiency and robustness of the proposed method of watermarking is checked using the SSIM (Structural Similarity Index). The images are computed in two situations: when the layers images are induced and on non-noise images. The Bland and Altman (B&A) plot indicates that both methods can be used successfully in securing the medical images.