Abstract:

A novel synthetic registration method is put forward. Based on alignment metric and normalized mutual information, image edge detection and image equalization are employed, and six registration algorithms are integrated to operate synthetic multimodal image registrations. Two criteria arethat the minority should subordinate to the majority and the weight be based on the credibility. Results of infrared image and the corresponding visible light image registrations are analyzed, which indicate that the synthetic registration method has high accuracy and robustness, and  has an obvious comparative advantage to non-synthetic registration methods.

Key words: multi-modal image registration; alignment metric; normalized mutual information; synthetic registration principles