Aircraft Targets Recogniton Research In Remote Sensing Images Based On The Features Fusion

Satellite remote sensing image recognition technology is a hot research topic in theworld today. With the development of science and technology, satellite remotesensing image resolution and continuously improve the quality, speed and shootingand transmission have been greatly improved, so that the target recognition technologybased on remote sensing image become a reality. Aircraft target as an importantstrategic resources, so the aircraft target recognition technology based on remotesensing image has wide application prospect. In the application of civil, can provide thedynamic supervision of civil aviation airport for flight conditions; in the military field, itcan provide real-time information to grasp the dynamic deployment and distributionof the enemy. But in fact, the remote sensing image has a large number of interferingfactors, such as noise, background, such as the weather, it will cause great influenceon aircraft target recognition.The paper begins with the introduction to the significance of aircraft targetsrecognition based on the high-definition satellite remote sensing images. Aircraftidentification as a kind of target recognition technology, which widely used in today’shigh-technology fields. Especially in the military field, fast, accurately identify theredeployment of his country’s military airport in all kinds of aircraft targets, canprovide information reference for analysis of the latest adjustment of militarydeployment, which won a key role for the victory of the war.In this thesis, aircraft recognition is the purpose of this study. By introducingvarious aircraft recognition algorithms in detail, this paper proposed the algorithm basedon the existing ones. The main content of this paper includes two aspects: First, aircraftrecognition algorithm; Second, the aircraft detection algorithm. The research of thistopic is from the two aspects, including the main ideas to improve the existing algorithmaccuracy for aircraft detection and recognition.In recent years, domestic and foreign aircraft target recognition technology hasmade great progress. At present, the recognition algorithm based on feature pointsand invariant moments is the majority, but there is no one can adapt to all kindsof aircraft targets. For the low precision, universal shortcomings in existing aircraftidentification algorithms, this paper puts forward the aircraft recognition algorithmbased on the feature points and invariant moments. The feature points can describe the objects’ precise information for distinguishing different aircraft types. The invariantmoments’ stability is good, and can keep good invariance with many kinds ofinterference. So the aircraft identification algorithm based on the feature points andinvariant moments has strong anti-interference and high accuracy.As to the slow speed and weak robustness in the existing aircraft recognitionalgorithms, this paper proposed the aircraft detection algorithm based on the saliencymap and invariant moments. Saliency map is the gray image calculated using specificsignificant theory, which can filter out a lot of background or interference parts, andhighlight significant goals in original image, so it’s conducive to the fast locatingcandidate targets. And invariant moments have the advantages of strong anti-interference, simple principle, convenient in calculation, easy to realize. Thus theaircraft detection algorithm based on the saliency map and invariant moments has theadvantage of less time spent and high robustness.Finally, this paper summarized the contents and ideas of this paper, then analyzedthe innovation highlights and shortcomings in this research, and showed the prospect offollowing research contents.