It has been established that the result of processing images obtained from on-board optical-electronic surveillance systems depends on the quality of the image segmentation method, which in turn poses the problem of the development of techniques and the selection of indicators for assessing the quality of image segmentation before the developers of imaging systems. The possibility of using a structured detector with the method of machine learning Random Forest for isolating boundaries on an image obtained from an on-board optical-electronic surveillance system is considered. An experimental study of the delineation of boundaries by a structured detector in an image obtained from an on-board optical-electronic surveillance system using Random Forest's computer training was carried out. The main advantages of using the method of machine learning Random Forest when allocating borders on the opto-electronic image are noted.
optical-electronic image, border allocation, method, machine learning, on-board system, detector, Random Forest, segmentation, decision tree, sampling, training