Robust and Fast Registration of Infrared and Visible Images for Electro-Optical Pod

Abstract

To deal with the registration of infrared and visible image with significant difference in contrast and large geometric distortion in electro-optical pod, a robust and fast registration method by using common structural features is proposed in this paper. First, to adjust the scales of input images, a geometric transformation model is built based on parameters of the cameras. Then, a global structural feature extraction strategy is developed by using phase congruency and significance ranking space. Those features are not only invariant to the scale and contrast but also embody the structural of the input images maximally and uniformly. Finally, to enhance the robustness to contrast, blurring, and incompleteness of structures, an adaptive feature matching method with kernelized correlation filter is presented. Experimental results demonstrate that the proposed method has better performance than several state-of-the-art methods in robustness and precision, and also confirm its validity in the real environment.