Visible and infrared image fusion using an efficient adaptive transition region extraction technique

Abstract In order to track a targeted environment, concealed weapon detection, navigation and military require various imaging modalities, for instance, visible image (VI) and infrared (IR) image. These modalities provide additional details. Complementary information from these images need to be fused into a single image for improved situational awareness. Hence, an ideal fused image should assimilate the essential bright information from the IR image and retain much of the original visual information from the VI. To achieve this, a region based image fusion technique using an efficient adaptive transition region extraction (ATRE) strategy is suggested in this paper. For the first time, the transition region extraction based approach is brought into the context of visible and infrared image fusion. This method is beneficial because it overcomes the problems of noise sensitivity, poor contrast and blurring effects associated with the conventional pixel-based methods. The proposed ATRE technique is used to efficiently extract the bright object regions from the IR image and retain much of the visual background regions from the VI. An adaptive parameter is introduced for accurate segmentation. A region mapping process is followed to get the fused image. Our technique is tested on standard fusion datasets. Image inspection and objective fusion indices are utilized to validate the results. They are compared with conventional and current pixel based and region based fusion techniques. The outcomes reveal that the suggested technique is comparable or better than state-of-the-art fusion techniques.