On the application of two-dimensional chirplets for resilient digital image watermarking

Chirplets (swept-frequency harmonic signals) find applications in several areas, including real and synthetic aperture radar, (Fresnel) optics and image processing. The robustness of chirplets to extreme (additive) noise make them an ideal choice in the role of embedding patterns for resilient digital signal and image watermarking. In this paper we present a new watermarking technique which is based on embedding twodimensional chirplet coded binary sequences in digital images. This approach provides a pattern embedded recovery algorithm which is robust to a variety of distortions. However, chirplets have certain parameters that need to be specified, and, given a set of target characteristics, some particular chirplet parameters can be considered optimal. This paper reports on the optimisation of both the chirplet sets and their associated parameters focusing on a classification of their resilience to image distortion.