The traditional detection of minimum resolvable temperature difference (MRTD) is mostly under laboratory conditions, and the results are obtained by the tester's subjective interpretation. Because of its large and complex system Structure, it is very difficult to apply in the field on-line detection. In this paper a set of field on-line detection system of infrared thermal imager is designed, single blackbody control technology is adopted to simulate temperature difference of four bars target and environment, and a coaxial Cassegrain collimator is used to simulate infinity target, through this way volume and quality of the detecting equipment is effectively reduced. Instead of human eyes we use CCD to capture the thermal imager's imaging result, an effective filtering method which combine linear filter and nonlinear filter is adopted to reduce noise and enhance the captured image, then four bars target area is accurately extracted as the target area, to reduce noise a rectangular area which just contains four bars target is extracted as the background area. Using relationship between temperature difference of four bars target and environment and gray level difference in corresponding infrared image, the parameter MRTD of the infrared thermal imager can be calculated, the experiments show that the result is in good agreement with the result of the human eye observation. The detection system has a simple structure, small size, low cost, high level of automation, etc. It can realize the on-line automatic detection of MRTD under field conditions better.
