Combining visual and IR images for sensor fusion: two approaches

This paper describes and compares two different methods for combining multisensor images into single integrated pictures for visual data analysis and data exploration. In the specific case considered here, the original images are thermal (IR) and visible. The first method preserves contrast in the thermal image and modulates local contrast by the structure of the high- frequency information in the visible image. This method produces a conventional gray-scale picture. The second method encodes the intensity at each pixel position in each image as the length of a line-segment, or 'limb,' of a stick-figure icon at the corresponding position in the output picture. This method produces an 'iconographic' picture. Although these two approaches differ significantly, they both satisfy the goal of incorporating the unique features of the thermal and visible images in a single integrated picture. We discuss the strengths and weaknesses of each method, and we suggest ways in which each might be improved and extended.