Detection of feature points on the three-dimensional scene under varying conditions of observation

In this article, authors proposed feature detection algorithm, which stable to the change in conditions of observation. Upon detection of the signal in element with unknown intensity level, another set of background elements used as additional samples with the information on noise (background) process. Equations for the calculation of the false alarm rate, correct detection probability and the threshold constants were developed. Simulation proved the stability of the decision rule used to detect a feature point in a discrete space with varying mean and standard deviation of the background radiation intensity. Simulation confirmed the possibility of applying the theoretical equations for correct calculation of the detector probability characteristics.