Automatic object detection in optoelectronic systems with a bolometric matrix receiver

Abstract

The paper considers an optical-electronic system (OES) of the infrared (IR) range, which implements an estimation-compensation approach to image processing to ensure a constant false alarms rate (CFAR processor) in an automatic object detection device. The primary image is formed using a micro bolometric matrix radiation receiver (MRR). The operation of the CFAR processor is based on a parametric algorithm for estimating the average value, variance and background gradient in special analysis windows surrounding the area of the probable appearance of the object. The relations for calculating the adaptive threshold for detection in the presence of Gaussian noise and correlated background are obtained. The results of experiments on image processing with an unsteady background and a point target are presented. When automatically detecting a target in the CFAR processor, two types of estimates were used: a sample average and based on extreme statistics. For these cases, the main features of the CFAR processor, its advantages and disadvantages are discussed.