A Study of Two-point Multi-section Non-uniformity Correction Auto Division Algorithm for Infrared Images

Traditional two-point multi-section non-uniformity correction method for infrared imaging use fixed number of sections and division points; this would greatly affect the algorithm's performance. In order to solve this problem, an auto division algorithm based on slope difference of average temperature response curves of IRFPA and least square fitting was supposed. This algorithm calculates the gradient of IRFPA's average temperature response curve and find out the turning points, then the approximate positions of division points are decided according to given number of sections or given threshold for slope difference; after that a least square fitting is performed in each section to produce subsection correction lines; finally adjacent lines are connected with each other to get the complete correction curve. Experimental results showed that the residual non-uniformity of an infrared image is greatly diminished from 0.08 % of traditional two-point multi-section method to 0.03 %. The method is still as simple as the traditional algorithm, while the precision of correction is significantly increased.