Neural Network Detection of Runway and Taxiway Marking Lines
摘要
This article presents the results of a study on the application of modern neural network methods for detecting runway and taxiway markings of an airfield in real time. The methods under consideration, originally developed for traffic lane detection tasks in automobile transport, are adapted for use in multispectral aircraft vision systems. The experiments were conducted on a laboratory stand simulating the operation of a machine vision system, using a model of an onboard neural network computer. The software part is based on modern neural network architectures, ensuring high quality detection and speed sufficient for real-time operation. To test the proposed approach, synthetic multispectral data obtained using computer modeling, as well as real data from open sources, were used. The results of the experiments confirm the effectiveness of using neural network methods to solve the problem of detecting runway and taxiway markings of the airfield in the conditions of modern air transport.