Research On Aerial Image Vectorization Technology And Its Application In Navigation

In modern air navigation field, scene matching aided navigation has been widely used in variouscountries’ aircraft and missile navigation systems. But the raster image used in scene matching systemhas huge data size so that the airborne database can only store a smaller geographic scope. Vector mapcan use a smaller amount of data to show a more complete ground features. Vector based mapmatching navigation has the advantage of high positioning accuracy, a wide navigation range, higherpositioning timeliness. A new way is provided to the aircraft navigation.Based on the program of the scene matching aided navigation system, combined with thecharacteristics of vector data, a vector map matching navigation system is proposed. The extraction ofwater target in aerial image, the vectorization algorithm for linear target and the matching algorithmfor vector map are studied to implement the vector map matching navigation system.Firstly, the structure and characteristics of the vector data will be studied. The navigation schemeof the exist mature scene matching navigation system will be analyzed. And a vector map matchingnavigation project will be studied. The combination of two will give full play to the advantages thatthe vector image concise expresses graphical information and has a small data. Next, the researchabout object extraction technology for real-time aerial image will be studied. Water area is selected tobe the extraction target of real-time aerial image. A two-step water extraction project is put forward toextract water target results which can be applied for navigation. Then, for the vectorization of lineartarget in raster images, a vectorizing algorithm based on special-point tracking will be studied, whichcan effectively achieve vectorization of linear target, include lines intersect case and closed curve case.At last, for vector map matching, based on the invariance of the curve’s difference angle characteristic,combined with the concept of Hausdorff distance, a vector matching algorithm based on angelHausdorff distance will be proposed and it solves the matching problem when real-time map containsmistaken vector, and finally give the relationship of rotating, panning and scale between the real-timemap and reference map.The results show that the proposed vector map matching navigation system is feasible and hasimportant practical significance and application prospects.