| With the development of the digital age, today’s medical images can be storedand transmitted through the network, and image tampering has become very common.So how to solve the authenticity and integrity authentication of the medical images isvery urgent. This paper combines the characteristics of medical image and digitalwatermarking technology, and proposes a watermarking algorithm for medicalimage in the field of anti-counterfeiting technology. The main contents are as follows:1. Hologram watermarking method based on DWT and SVD is proposed in thispaper. According to the characteristic of not being tore up for the holographicwatermark and singular value decomposition, this method gives improvements inrobustness to various attacks. Firstly, a watermark image is used to create a hologram.Then the medical image is decomposed by secondary wavelet transform and singularvalue decomposition. The frequency components are embedded into the intermediatefrequency components. Experimental results showed that the algorithm has goodinvisibility and is robust against noise, filtering, translation, scaling, JPEGcompression, especially, watermarking can be detected after print-scan.2. This paper proposes a watermark algorithm for medical image based on linearenhancement hologram. The algorithm is designed to enhance the robustness by usingthe characteristics of a linear enhancement hologram. First, the watermarkinformation is encoded into a QR code image, and converted it to the binary QR codepattern. Second, the QR code image is created as a hologram. Then, the hologram islinear enhanced to be the watermark. Finally, the watermark can be embedded into thesubband of the discrete wavelet transform of the medical image. In the process ofwatermark extraction, to improve the accuracy of decoding the watermarkinformation, we use the compensation error correction method of QR code.Experimental results showed that the algorithm can solve the watermark contradictionbetween the robustness and invisibility, and can resist common attacks, it has strongrobustness.3. Concerning the low accuracy of the present authentication algorithm, asemi-fragile watermark algorithm for medical image has been put forward. Thealgorithm puts forward the similarity of watermark, it greatly reduces the false-alarm probability. Firstly, original image is separated into blocks, then to calculate the high5-bit gray-scale for the mean of each pixel as the image characteristics watermarkinformation, and embed it in the two least significant bit of the host image. Accordingto the method of parity to recalculate the watermark as authentication watermark, andembed it in the third from bottom in the significant bit. Finally compare the similaritybetween the authentication watermark and the extracted watermark. When thesimilarity is higher than0.6, the certification is correct, otherwise, it is wrong. Tamperpositioning accuracy of the algorithm is2×2pixels in an image block. Theexperimental results verified the accuracy, safety and practicality of the algorithm. |
