Research On Key Points Detection Of Air Non-cooperative Reconnaissance Balloon

Detecting and striking the critical part of small aircraft is an important way to counterattack the unmanned aerial vehicle.The Reconnaissance balloon is one of the airborne vehicles and is a widely used high-altitude reconnaissance auxiliary equipment.The detection equipment is generally suspended at the tail of the balloon.The illegal use and abuse of reconnaissance balloons poses a serious threat to citizens’ privacy and even national security,so they should be countered.The accuracy and real-time of the detection of the key points of the detection balloon are difficult to solve in the existing image target detection and tracking methods.Aiming at the above problems,this paper proposes a balloon detection and tracking algorithm based on curvature feature in polar coordinate system.In order to save the detailed information of the image,the algorithm firstly uses the bilateral filtering to preprocess the infrared image.The processed noise is effectively suppressed,and the edge information of the balloon is well preserved.In order to locate the approximate position of the balloon,this paper designs a The third-order backward differential template successively locates the balloon through the horizontal and vertical projection methods and the third-order backward differential template operation,and obtains the approximate center point of the balloon.For the problem that the curvature of the balloon boundary is weak,this paper takes the approximate center point of the balloon as the The origin of the polar coordinates converts the balloon to the polar coordinate domain to achieve the effect of enhancing the boundary curvature feature.Then,in order to meet the real-time requirements as much as possible,an improved fast Otsu algorithm is proposed to segment the image and extract the boundary of the balloon.Line;finally,the least square method is used to obtain the curvature characteristics of the boundary line,and the prior information is used to verify the key points.In this paper,the proposed algorithm is compared with SIFT and KCF algorithms.The two sets of infrared balloon sequence images are selected as test sets.The effectiveness of the proposed algorithm is verified by quantitative and qualitative analysis methods.The algorithm is implemented on the DSP(TMS320C6678)system and optimized by various optimization methods.By comparing the time-consuming beforeand after optimization,it is shown that the optimization of the algorithm can greatly reduce the time-consuming,and meet the requirements of real-time detection on the DSP system.In summary,it shows that the algorithm can be used in real-time detection on the DSP(TMS320C6678)system,and has high accuracy.It has important value and significance for practical application.