The Researching About Detecting Methods Of Foreign Object Based On The Mobile Robot Vision

With the rapid development of social economy, the arrival of information age as well aslife quality increased ceaselessly, people pay more attention to traffic safety, the status of testand evaluation of traffic safety has been one of the key factors. The requirements of theprecision in the foreign object detection system on airport road become much higher,especially in the rapidly developed aviation sector.The algorithm of detection of mobile target is researched in this article, and taking therunway as the background, proved with the mobile robot testing-platform, lay the foundationfor further development of the practical application system. Firstly, a visual system suitablefor this topic is determined by comparing different visual systems imaging model anddiscussed the composition and classification of the mobile robot vision system; Secondly, amethod of combining frame differential and SIFT algorithm is proposed through thecomprehensive analysis of various common target detection algorithm and the way of thecommon foreign body detection. The principle of frame differential method is simple, it canovercome the light intensity, the weather changes, the changes of external circumstances andgains pretty ability to resist disturbs. But the accuracy of the moving target in the road foreignobject detection is not quite high, easy brings empty phenomenon. Lots number of featurespoint are extracted with the SIFT algorithm, it can partition one or more goals according tothe different characteristics of different foreign information, but the computation iscomplexity, slower the operation speed. So a new algorithm combining the above twoalgorithm has been proposed, make up the flaws of each one and guarantee the accuracy aswell as real-time of the whole system; Finally, the foreign object detecting system on therobot is designed using the platform of mobile robot in lab, based on the mobile robotplatform. The experiment has been done adopting the frame differential method, algorithm ofSIFT, and the fusion of frame differential plus SIFT algorithm. The result of the experimentalproved that the method raised in this paper is better than single algorithm of frame differentialor SIFT algorithm. The experimental verify the detection result through the method ofcombining frame differential method and SIFT algorithm on the road of foreign objectdetection is pretty good.