Methods of formation and processing of infrared images

Problem statement. Effective use of IR sensors is possible if it is known what an aerial object will look like in the IR range (the presence of a reference image). Computer modeling is currently widely used to create such reference images. Based on the information about point objects contained in the image, it is very difficult to achieve a high probability of detecting point objects while maintaining a low frequency of false interference because point targets have too few reliable signs to distinguish them from small objects on the ground. The article discusses the task and methodology of detecting point, moving and stationary objects both on the Ground and in the air using equipment with infrared sensors, processing digitized information, image quality improvement methods, target detection against a background of non-stationary interference. The purpose of the work is to develop in the form of software the principles of organizing the processing of sequences of digitized IR images in the interests of reliable detection of small (described by no more than five pixels) objects that are grouped, creating a formation, as well as the development of devices to eliminate noise while maintaining the shape of objects, improve image quality, detect objects against a background of non-stationary interference, suppression of broadband pulse interference. Results. When building detection algorithms, considerable attention is paid to the influence of electromagnetic interference on the images generated by IR sensors, as well as to the development, comparison and verification of methods for improving the quality of these images. The main goal of algorithms should be to eliminate noise while maintaining the shape of the targets. This will increase the efficiency of the algorithms considered at the stages of target detection, classification and tracking, as well as the formation of target designations. The best estimate is achieved when the operation is performed on an image in which noise has already been eliminated. Practical significance. Working in conditions of reflections from local objects requires the introduction of high-speed adaptive threshold devices into the signal processing path, which ensure the stabilization of the probability of a false alarm. Thus, the obtained values of the threshold constants make it possible to use a false alarm probability stabilization device (SVLT), which uses the method of ordinal statistics in image processing and thereby ensure the normal operation of the device under the influence of non-stationary interference.