Calibration of thermal cameras with hot patterned objects

Thermal cameras are great if you want an idea of ​​what’s hot and what’s not. However, if you want to use a thermal camera for certain machine vision tasks, you generally need to perform a geometric calibration to understand what the camera is seeing and correct for lens distortion. (Henry Zhang) has shared various methods to do just that.

To calibrate a thermal camera, you must first have a thermal pattern. This is like a typical test image for a camera or monitor, but with temperatures instead of colors. (Henry) explains several methods of doing this. One involves using a grid of nichrome wires to create a thermal pattern for calibration purposes. Another uses discs of cold aluminum inserted into a foam board. Even a simple checkerboard can work, with the black spaces being heated more from ambient sunlight than their adjacent white spots. (Henry) then explains the mathematical techniques used for calibration based on these patterns.

It is a useful primer on the subject if you work with thermal camera systems. We’ve also looked at some other interesting machine vision topics before. If you have any great thermal imaging tips, don’t hesitate to drop us a line!