A Neural Network Based Compensation of Thermal Infrared Data Considering Environmental Temperature Variations

Abstract

In this paper, infrared data correction algorithm is suggested based on neural network functional approximation when the environmental temperatures such as air temperature are varying. To compensate infrared data, the relationship between infrared data and environmental temperatures is investigated first. Based on the relationship, a neural network approach is applied to identify a function which can be utilized to compensate the influence of environmental temperatures on infrared data. Through experiments the proposed neural network based algorithm is shown to reduce the influence of environmental temperatures on the infrared data effectively by comparing with polynomial based functional approximation approachs