Copy-move is one of the most common method for image tampering. Many methods have been proposed to detect the tampered regions. However, it is hard to detect the tampered regions when they are rotated before being pasted. To solve the problem, this paper presents a novel approach for detecting region-duplication forgery with rotation. Polar Harmonic Transform (PHT) is employed to extract the features of the circular blocks, which are then used to perform block matching. Experimental results show that the proposed method is not only robust to added noise and JPEG compression, but also robust to object rotation. Simulation results demonstrate the performances of the proposed scheme.