Abstract: The robustness against geometrical transforms is still one of major concerns in robust watermarking although significant improvements have been achieved in past decades. This paper tackles the global geometrical attacks by designing a deformable multi-scale transform (DMST) with joint shiftability in position, orientation and scale. Via DMST, we derive theoretically the principles for geometrical synchronization and then utilize it to develop a template-based scheme to efficiently estimate geometrical parameters. The hidden Markov model in the standard wavelet domain is extended to the steerable wavelet domain, which is further used to improve the watermark extraction performance. Experimental simulation demonstrates that the proposed watermarking scheme is quite robust to common signal processing, global geometrical attacks, and their joint attacks.