Determination of Wavelengths That Make Infrared Guided Missile Easier to Differentiate From The Background

This study is carried out to determine the bands that allow the guided missiles to be separated from the background more easily. Pixel samples collected from hyperspectral background images and synthetically produced missile images were used as reference dataset. The AdaBoost classifier is trained with the pixel samples which are dimensionality reduced. The indices of the correctly classified samples were kept during the evaluation process. After the indices were determined, the reduced dimensionality of the data was converted back to its original dimensions. By using correctly labeled samples, the total radiance values were found for both the corresponding class and the other classes. Then, the average of the total radiance values was taken by using the lengths of the correctly labeled samples. After finding the average radiance values, these values were compared to find the wavelengths that separate the IR guided missile from the background.