Parallel implementation of machine vision for aircraft navigation

Abstract

So called, correlational extreme navigation systems provide extra accuracy due to corrections made from external sources. In this work, a machine vision based navigation for low altitude air operation is considered. An approach to navigation based on utilising time-invariant features of aerial images, such as linear features of artificial or natural origin (roads, areal feature contours) is presented. Though, there are many works in the area of remote sensing about linear features detection, only a few of them consider "live" usage on board for service needs. Some theoretical aspects of using different navigation features considered, the algorithm and results of the processing are presented. Recommendations for future research are proposed.