Research On Medical Image Of Reversible Data Hiding Algorithm Based On Code Division Multiplexing

In recent years,the illegal collection and modification of medical images have gradually become a high-profile issue.Most of the sensitive information of these multimedia is easily exposed to attackers in an open wireless network environment.Therefore,in order to guarantee the security of multimedia data such as medical images,they need to be carefully protected.Reversible data hiding is a special method that can embed secret information into multimedia.Besides,the secret information can be recovered from the encrypted image without loss after completely extracting.At present,the traditional multimedia protection,especially the reversible information hiding method of medical images faces many challenges.Because medical images are very sensitive,even small changes in pixels may have a huge impact on disease diagnosis.If the medical images changed slightly,the diagnosis of doctors may be affected therefore.The main content of this paper is: In the first part,a reversible data hiding algorithm for medical images based on code division multiplexing and machine learning is proposed.In the process of embedding information,this method applis two layers of small sample neural network to optimize the embedding strength,sequence length and other parameters,so as to obtain better peak signal-to-noise ratio and structural similarity.According to the orthogonality characteristics of different extended sequences in the code division multiplexing algorithm,secret information can be repeatedly embedded in medical images,and most elements of the extended sequences can cancel each other,effectively improving the embedding ability and security of reversible data hiding.In the second part,a real-time image reversible data hiding scheme based on code division multiplexing is proposed.In this scheme,medical images are divided into foreground and background areas.The medical images are divided into blocks and sorted according to the mean square error.The integer wavelet transform algorithm is used to convert the original medical images into the frequency domain.The patient's private information is embedded in the high-frequencysubbands of medical images.Therefore,the proposed method has a low image distortion rate even when the embedding capacity is high.In the third part,we present a reversible data hiding method with multi-directional gradient predictor.The multi-directional gradient prediction method is used to embed secret information into the cover image,and the gradient information of the pixels in four directions is calculated to obtain smaller prediction errors and obtain more accurate prediction values.Compared with some out-of-state schemes,the proposed algorithm has a higher peak signal-to-noise ratio than traditional prediction error methods.