Visible-Infrared Image Discerner (VIID)
Computing & Wireless : Application Software
Available for licensing
- Alan Bovik, Ph.D. , Electrical and Computer Engineering
- Todd Goodall , Electrical and Computer Engineering
Thermal imagers capture a wide variety of content, but the resulting thermal image or video can widely vary in quality. Knowing the statistical differences between infrared and visible light imagery can allow better quality control within the imager.
Visible-Infrared Image Discerner (VIID) provides the capability to distinguish infrared and visible light natural image, or video statistics, while simultaneously providing a quality measurement on the captured image or video, regardless of content. Prior to the development of VIID, no method or algorithm existed which could distinguish between infrared and visible light images or videos. Additionally, no software existed which could evaluate quality on-the-fly in a perceptually relevant way.
- Requires little training
- Has a high classification accuracy
- VIID is distance based, so a thermal imager can place thresholds on the distance to capture an optimal infrared image or video.
- A dual infrared/visible light camera setup can use distances provided by VIID to refine camera tuning parameters.
- VIID is written entirely in Python, providing cross-platform support.
The discerner was developed as part of infrared image/video quality assessment research, and as such, can have many quality-based research applications. IR manufacturers can use VIID when coupling infrared and visible light cameras and for in-the-loop quality control.
- 1 U.S. patent application filed