A new dct filters-based method to improve the resistance of ternary watermarks in audio files against attacks

Abstract

Watermarks used to verify the authorship of various works of art is a simple tool available to novice authors. This paper proposes a new technique to enhance the resistance of watermarks in audio files. As parts of a melody or image, leveraged watermarks can be recognized by human beings, even if dramatically corrupted. Nevertheless, increasing the quality of extracted watermarks is always helpful. This is achieved by new methods developed for damaged watermark restoration. In our methods, the source media is converted into a ternary code. The suggested technique is based on the restoration of a part of the source’s features via its ternary code. Discrete cosine transform matrix rows are used as the final impulse response fil-ters. Some features of the source are extracted by applying a linear combination of these filters to the ternary watermark. We consider the two most frequently occurring attacks: filtering and converting into mp3 format.