LongWave Infrared and Visible Image Fusion for Situational
Long-Wave Infrared and Visible Image Fusion for Situational Awareness Nathaniel Walker
Agenda • What is image fusion? • Applications • System-level considerations • Image fusion algorithms • Image quality metrics • Further research
What is Image Fusion? • Combine data from multiple sensors into a single image • • Visible Image Intensified (I 2) Near Infrared (NIR) Short-wave Infrared (SWIR) Medium-wave Infrared (MWIR) Long-wave Infrared (LWIR) X-Ray • Enhance the capabilities of the human visual system • ‘See’ outside the visible spectrum • All-weather visibility
Applications • Surveillance and Targeting • Satellites • Navigation • Guidance/Detection Systems
System-Level Considerations • • • Parallax Optical alignment Image registration Sensor pixel resolution Color vs. grayscale • Spectral resolution can be lost in fusion • Human factors • Presentation of IR data • Realism of displayed data (superposition, contrast reversal) • Preserving relative intensity across the scene
Image Fusion Algorithms (Zhang, Blum 1999) • Weight-based combinations of the two sources • linear combination • general loss of contrast • Feature extraction • High-pass filtering or edge detection • Maximizing image quality metrics
Image Quality Metrics • Mostly done by subjective evaluation • ‘Optimal’ methods are task and application dependent • Two classes of quantitative metrics (Chen, et al. 2005) • Analysis of the fused image • standard deviation – measure of contrast • entropy - measure of information content • SNR • Comparison of the fused image to the source images • cross-entropy • objective edge based measure • universal index based measure
Further Research • Concentration on grayscale fusion algorithms for effective communication of spectral information to the viewer • Sensor Assumptions • perfect optical alignment and image registration • same pixel resolution and field of view (FOV) • Compare quantitative metrics of image quality to subjective image evaluation for situational awareness • Focus on human factors for injecting infrared content into a visible spectrum image • What approach adds value without causing distraction or removing detail
References
- Slides: 9