Hybrid NSCT domain multiple watermarking for medical images

Abstract

Medical image watermarking is a challenging area of research. High bandwidth, secure transmission of patient's data among hospitals and hiding capacity are major concerns in medical image watermarking. Recently, wavelet transforms and their hybrid combinations have been widely used for this purpose. The conventional wavelet transforms suffer from shift sensitivity and have low hiding capacity. Therefore, the performance of hybrid combinations of wavelet transforms for image watermarking is limited. The shortcomings of wavelet transforms can be overcome with the use of Nonsubsampled contourlet transform (NSCT) which is shift invariant in nature and provides rich directional information. For these reasons, in this work we propose NSCT based technique for medical image watermarking which combines discrete cosine transform (DCT) along with Multiresolution Singular value decomposition (MSVD) and Arnold transform in order to increase robustness, capacity and imperceptibility. In the proposed work, multiple (three) image watermarks have been used for a single cover medical image. We have embedded three image watermarks into NSCT coefficients of the cover image. Among which two of them are image watermarks and third is encrypted text watermark. By using NSCT, embedding capacity has been increased and it becomes more resistant to geometrical attacks. Also, hybrid combination of NSCT with DCT, MSVD and Arnold transform increases the perceptual quality and security of watermarked image. Experimental results showed that the proposed method provides high robustness against attacks like JPEG compression, Rotation, Resizing, noise, Blurring and found better than existing hybrid methods for medical image watermarking.