An embedding technique for smartcard-supported e-healthcare services

Abstract

Generally, a smartcard contains the photograph of its owner. Few recent applications manage patient’s information by embedding them into that photograph. The historical records of a patient may increase gradually. The volume of raw data onto growing records could be unmanageable by a scheme. Therefore, an objective of this research is to encode every raw data by 1–3 bits owing to minimize the length of the embedding information. Prediction error-based reversible embedding schemes present higher embedding capacity and image quality. In the prediction error space, it is investigated that the frequency of − 2 is noticeably smaller than − 1, similarly, − 3 than − 2 and so on. The scenario is valid for positive-valued errors as well. For this, every single-layer embedding scheme implants bits in the errors of − 1 and 0 only. Another investigation reveals that secret data of a patient with chronic diseases may exceed the frequency of − 1 and 0 valued errors. Hence, another objective of this research is to implant bits in the errors of − 1 and 0 for multiple times. Considering these objectives, we have proposed a model that encodes raw data of patient’s history and embeds the encoded data in an image for multi-times. The encoding technique both minimizes the length of embedding information by 1/10th and enhances the security of the implanted data. The proposed model is verified by experimenting with MATLAB 7.0 on BOSS and CalTech 101 image datasets. Experimental results present a double gain in the embedding capacity for the proposed scheme. It will be a notable contribution to the field of medical applications.