Framework for audio reversible watermarking robust against content replacement with signal restoration capabilities

Abstract

In the literature, there are robust Reversible Watermarking Schemes (RWS) that allow the extraction of watermarks after the signals have suffered attacks, however the modified carrier signals cannot be restored. On the other hand, self-recovery schemes allow the restoration of the attacked regions of the signals, but the embedding capacity is used for control information, and secret messages cannot be inserted in these type of schemes. In this manuscript a framework for robust reversible watermarking with signal restoration capabilities is proposed. This framework is a general construction that can be followed to implement schemes for different type of media. Previous works that propose a RWS and a self-recovery scheme, both for audio signals, have been taken to construct an implementation of the framework for audio signals. These schemes use the auditory masking properties of audio segments to insert the secret message and the control information required for restoration in a transparent manner. Results obtained with this implementation demonstrate that this construction fulfills a transparency constraint of −2 Objective Difference Grade (ODG) for practical applications, and it reaches payload capacities up to 13 kilobits per second (kbps) with acceptable robustness and adequate quality for the restored signals.