On an Information Embedding Method Using Redundancy of Prediction Selection in Adaptive Predictive Coding

Abstract

There have been increasing chances to exchange image data. In order to handle or manage images easily, it is desirable to embed the attribute information such as copyright ID of the image into the image itself. In our preceding study, Sugeno et al. proposed a method to incorporate data into an adaptive lossless coding method. In our method, coding parameters such as prediction functions are controlled by the embedding data and the original image data are never changed. However the method to expand the range of the selective prediction functions was never considered. This thesis addresses a coding method for embedding information, which improve the compression ratio by increasing the nember of candidate prediction functions . The conventional embedding method by block-based adaptive predictive coding uses two prediction functions in embedding information. However, depending on the statistical property of the block, the same compression ratio can be obtained by the other prediction functions. In this case, utilizing these functions can increase embeddable information than in the conventional method. Simulation results show that the method of utilizing these functions has a little improvement in some cases compared with the conventional adaptive predictive coding.