Infrared and visible image fusion based on region growing and contourlet transform

According to the characteristics of infrared and visible images, a new image fusion method based on region growing and contourlet transform is proposed in this paper. To obtain more complementary information, the method is designed as a two-stage procedure. Firstly, the input infrared image is processed with region growing to segment the thermal target. Different fusion rules are adopted in target and background regions, respectively. For the target region, local energy is utilized as the fusion rule of the first fusion to fuse the thermal target and the visible image, while for the non-target region, we reserve the visible background information. Secondly, in order to fully add original information of the source images and avoid loss of information caused by segmentation, we make the second fusion between the visible image and the result image of the one-stage fusion. For good properties of localization, directionality and anisotropy, we adopt contourlet transform as the second fusion method. Experiments are carried out and the results show that our method is clearer in visual quality and effective in quantitative evaluations and the fused images are better than those resulting of using wavelet transform and contourlet transform.