Fusion of infrared and visible images based on focus measure operators in the curvelet domain

Aiming at the differences of physical characteristics between infrared sensors and visible ones, we introduce the focus measure operators into the curvelet domain in order to propose a novel image fusion method. First, the fast discrete curvelet transform is performed on the original images to obtain the coefficient subbands in different scales and various directions, and the focus measure values are calculated in each coefficient subband. Then, the local variance weighted strategy is employed to the low-frequency coefficient subbands for the purpose of maintaining the low-frequency information of the infrared image and adding the low-frequency features of the visible image to the fused image; meanwhile, the fourth-order correlation coefficient match strategy is performed to the high-frequency coefficient subbands to select the suitable high-frequency information. Finally, the fused image can be obtained through the inverse curvelet transform. The practical experiments indicate that the presented method can integrate more useful information from the original images, and the fusion performance is proved to be much better than the traditional methods based on the wavelet, curvelet, and pyramids.