Development of a Thermal/Visible Image Fusion System for Situation Awareness

Abstract

The development of an image fusion system is described in this paper. The system consists of thermal and visible (color) cameras. Image alignment between two camera views is performed by using homography. The homography was estimated from four pairs of corresponding points on a planar object in the captured images that was manually selected by the user. Then, the user-defined region of interest (ROI) is used to adjust scale and position of the two images. Finally, the image fusion algorithm is applied to create the fused image. By thresholding the thermal image at a specified intensity level, the pixel values in the color image will be replaced with a predefined color if the corresponding pixels in the thermal image exceed the threshold value. This technique can provide thermal information of possible targets to be fused with original color image. This makes the system suitable for situation awareness. The experimental results show that the system can operate in real-time with satisfactory results.