الفهرس | Only 14 pages are availabe for public view |
Abstract The security of images or les transferred via internet is very important issue. Especially the patient information les at hospitals. These images must be secured before sending it through internet. The requirements of securing the image that transferred via internet are condentiality, integrity, source authentication and availability. The proposed model satised most of security requirements to ensure that any medical image related to any patient can not be accessed via any unauthorized person. Applying encryption technique in the medical image to ensure that only authorized person can decrypt this medical image and can obtain the original image. By using encryption the proposed model ensured authentication and integrity. The ownership of these medical images is very important to prove. The ownership of any medical image is identied by using watermark related to the owner of this medi- cal image. The name of patient and serial number is used as watermark, the ear image is captured and used it to identify ownership by extracting features of patient’s ear image and embedded it into the original image. It’s very important to secure these features because it includes important patient information (ear image). Condentiality and own- ership are ensured via encrypted information related to the patient and embedded it into medical image. Traditional methods for personal identication are based on what is the person knowing like PINs (Personal Identication Numbers), log-ins, passwords, identication cards and specic keys. This method has a lot of disadvantages such as dicult to remember, any person can crack passwords, cards and keys are often stolen. Because of disadvantage of traditional methods for identication preferred to use biometrics. Ear biometrics is used because it has many advantages of using it as a source of data for human identication. Firstly, ears do not change considerably during human life. Secondary, Ears have both reliable and robust features which are extractable from a distance. Thirdly, Ear prints could be printed at a scene of crime. The size of the medical image is also very eective point in transmitting it via internet. If the size of the image is small, ensured good and fast transmitting. But if the size of the medical is large, the image will face many problems in transmitting such a delay. Because of this the proposed model suggested compression technique applied on medical image before sending via internet. On the other hand the run time of program is a very important aspects of computer science. To reduce run time of the proposed model, DWT(Discrete Wavelet Transform) is used to separate an image into four sub-bands (approximation image LL (Low Low), Horizontal detail HL(High Low), Vertical detail LH (Low High) and diagonal detail HH (High High)). LL contains the most structure of the image. Encryption technique is applied in the approximation image because it is the most important part in the image instead of applying it in all image. Also watermark is embedded into the least important part in the image to reduce the dierence between the watermarked image and the original. To return four sub-bands of the image (LL-LH-HL-HH) to one image, the proposed model applied (IDWT)inverse discrete wavelet transform. Then compression technique is applied to this image to obtain the less size of the image then transmit it via internet. Finally, various metric is used to evaluate the proposed model in each part sep- arately. In ear print used two algorithms, rst one depended on geometrical features and the experimental results showed that the algorithm obtained over all accuracy al- most 98% on IIT (Indian Institute of Technology) delphi database. Second algorithm depended on SIFT (Scale Invariant Feature Transform)features and the experimental results showed that the algorithm obtained over all accuracy almost 95.2% on IIT del- phi database, 100% on AMI(Analysis Mathematical of Images) database and achieve accuracy equal to 76%, 85% and 100% on the created database. The variation of the results depend on the size of training images. Encryption technique is evaluated by using four measuring error and all give free error between original and decrypted image under no active attacks. Two metric is used to evaluate similarity (NCC (Normalized Cross Correlation) and SC (Structural Content)). PSNR (Peak Signal to Noise Ratio) is used to evaluate dispersion of the encrypted image = 9:9 db. Watermark is evaluated by using four metrics (MSE (Mean Squared Error) between the original image and water- marked image= 1:3348 , SNR (Signal to Noise Ratio)= 24:0654, PSNR=96.5997 db, NCC=1). |