Research On Detection And Tracking Of Ground Moving Vehicles In Infrared Image

In the modern defense military system,infrared small target detection and tracking technology occupies a very important position.The detection and tracking technology of ground mobile vehicle has been widely used in missile guidance and satellite tracking system.However,due to the poor imaging quality of infrared image,the existence of noise,and the relative weakness of vehicle targets in high altitude perspective,the lack of features and other factors,there are still many difficulties to achieve accurate and robust detection and tracking of ground moving vehicles in infrared image.The existing detection algorithm based on filter is poorly robust even though it is fast,so it cannot be applied to complex scenes.On the other hand,optimization based detection algorithm perform inferior on time consumption in comparison with superior robustness,so it is difficult to implement in engineering.The existing multi-target tracking algorithms usually only consider the location information in target matching,which leads to the instability of the tracking process.The main contents of this paper are as follows:By analyzing the imaging characteristics of ground vehicle target and background in high altitude view,a fast two-stage ground moving vehicle detection algorithm is proposed.In the first stage,the local candidate regions with high probability of target occurrence can be located quickly through the multi-directional gradient fusion calculation of infrared image.In the second stage,a double-layer contrast detection algorithm is presented in local image area according to the target distribution characteristics,aiming to achieve accurate target positioning and eliminate false alarm.Through two-stage processing,the detection algorithm can achieve high accuracy and high timeliness.Compared with the traditional detection algorithm,the algorithm in this paper can not only realize fast detection,but also adapt to the strong noise,complex and changeable detection environment,showing the potential of engineering application.In order to solve the problem of insufficient information in infrared image tracking,a target tracking algorithm based on joint multi-feature matching is proposed.Firstly,position matching is carried out according to the motion features of the target.Then,feature matching is conducted according to the shape of the local area where the target is located.Compared with the traditional tracking algorithm,the joint multi-feature matching algorithm can track the target accurately and stably when the detection platform has a large jitter.Finally,in order to verify the engineering practicability of the algorithm,this paper implements the algorithm under DSP development platform.Through testing the detection and tracking algorithm,it proves that the algorithm has good application value in practical engineering.