| In the airport security system,the airport perimeter effectively guarantees the security inside the airport.At present,manual inspection is the main method.Therefore,using robots for the airport intrusion prevention and inspection can reduce the labor pressure of staff and improve the inspection efficiency.This paper designs and develops an airport fence inspection robot system from the perspective of mobile visual detection.The main work is as follows:Firstly,the overall design of the robot system is introduced,the hardware of the orbital robot is carried out,and designed the overall architecture of the upper computer software.Secondly,a method of moving target detection in the dynamic background is proposed.This method establishes a motion parameter model for the camera platform of a mobile robot,and then uses SURF algorithm to extract feature points and make feature matching.Then performs global motion estimation,transforms the dynamic background into a static background.Finally,uses Vibe algorithm to detect the moving targrts.Aiming at the problem of ghost and grid noise interference in detection,the construction process of the initial background model set was improved,and the fixed threshold was changed to the adaptive threshold which improved the accuracy and anti-interference ability of the algorithm.Then,a target tracking method based on the combination of particle filter framework and Camshift algorithm is proposed.On the basis of particle filter,the Camshift principle is used to iterate each particle,so that particles can quickly drift to the area where the target is,and achieve the goal of target tracking.Based on this algorithm,the tracking feature of the target is improved,and a target observation feature model combining edge features and H-S color features is proposed,which solves the interference problem of the tracking effrct when the obstacles are similar to the target color.The method has a moderate amount of calculation and good tracking effect,and has good real-time performance and reliability.Finally,based on the algorithms and system functional requirements,an upper computer system is designed.The upper computer system is designed based on Qt software,and each function interface is tested to verify the effectiveness of the visual detection algorithm.Through the whole system debugging,it is verified that the airport fence inspection robot system designed in this paper is real-time and feasiblility. |