Traditional temperature measurement algorithms for infrared thermal imagers have a common shortcoming of low measurement accuracy. The temperature values provided by them can not represent the real temperature of an object. In this paper, the temperature measurement principle of an infrared thermal imager is presented. The cause of the above shortcoming is given and the deficiency of the traditional temperature measurement algorithm using blackbody calibration is analyzed. Finally, an accurate temperature measurement algorithm for infrared thermal imagers is proposed. By incorporating the temperature difference look-up table calibration, temperature pretreatment and real temperature conversion into the traditional blackbody calibration algorithm, the new algorithm improves the temperature measurement accuracy of the infrared thermal imager.