A Colour-Focused Visible and Infrared Image Fusion Framework for Aiding Human Perception

Image fusion methods are required when images from the same or different sensors are available. In several applications, photographs from a colour camera are often fused with thermal infrared (medium-wave or long-wave) images. This paper focuses on those applications (typically security and surveillance) wherein the fused images are used by a human operator who will, in turn, make critical decisions. This paper presents two frameworks for fusing colour and infrared images, one in the red-green-blue (RGB) colour space and the other in lightness-chroma-hue (LCH) colour space. We compare qualitatively and quantitatively the performance of five existing fusion methods that work best in this scenario. The study is carried out across two datasets. Quantitatively, four existing metrics are used to analyse the performance. In addition, this paper proposes a performance metric that appeals to human perception and measures how well colour, edge, and contrast information are transferred to the fused image.