Real-Time Infrared and Visible Image Fusion System and Fusion Image Evaluation

Abstract

A real-time infrared and visible image fusion system is developed. This system consists of detector system, image processing system, transmission system, display system and control system. Image processing system is based on FPGA and DSP with the function of image fusion. A "contrast stretching" grayscale fusion method and the "color transfer" color fusion method in YCrCb space is applied in this system. To evaluate both grayscale and color fusion images, a color variety metric is proposed. This metric is based on color difference gradient in CIELAB color space. Two experiments were carried out using the developed system day and night. The experimental results prove that: first, this system could improve target detection because fusion images have more information than source images. Secondly, the proposed metric is meaningful and effective since it correspond well with human perception.