Researches on New Non-uniformity Correction Algorithm for Resistor Array

摘要

Resistor array is the main direction of the infrared scene generator development, its non-uniformity characteristic directly affects the quality of infrared imaging. The new correction algorithm based on neural network architecture, has better convergence speed and approximation accuracy bying choosing a suitable activation function, hidden nodes and the objective function. The algorithm is well compensated non-uniformity response of resistor array. The correction methods of off-line calibration and on-line look-up table can take full advantage of powerful computing capability, while it does not affect the real time of the infrared scene generation system.