A novel adaptive leak diagnosis and localization method for infrared image

Abstract

In this paper, a new adaptive diagnosis and localization method is given for small target leak points in an infrared image. The infrared image of a test vessel with cold compressed air is photographed by a thermal camera. Based on a mathematical model of the infrared image, the diagnosis and localization method is given as an adaptive algorithm which is synthesized by a developed adaptive center-weighted median filter and an improved adaptive kernel regression method. Experimental results are given to illustrate the proposed algorithm is effective and adaptable to small target detection under the complex infrared image background.