An adaptive method for the enhanced fusion of low-light visible and uncooled thermal infrared imagery

Abstract

Night vision sensors, such as image-intensifier (II) tubes in night vision goggles and forward looking infrared sensors (FLIR) are routinely used by U.S. naval personnel for night operations. The quality of imagery from these devices however, can be extremely poor. Since these sensors exploit different regions of the electromagnetic spectrum, the information they provide is often complimentary, and therefore, improvements are possible with the enhancement and subsequent fusion of this information into a single presentation. Such processing can maximize scene content by incorporating information from both images as well as increase contrast and dynamic range. This thesis introduces a new algorithm, which produces such an enhanced/fused image. It performs adaptive enhancement of both the low-light visible (II) and thermal infrared imagery (IR) inputs, followed by a data fusion for combining the two images into a composite image. The methodology for visual testing of the algorithm for comparison of fused and original II and IR imagery is also presented and a discussion of the results is included. Tests confirmed that the fusion algorithm resulted in significant improvement over either single-band image