Design Of The Tracking Algorithm Of Moving Target In Automatic Theodolite Tracking And Measuring

Automatic theodolites are widely used in aviation, military, and life. With the deve- lopment of automatic theodolite, cost-effective automatic theodolite, and automatic theodolite for a specific area become the focus of study. In this paper, aim to automatic theodolite used to track weather balloon, design and experiment the image processing algorithms of automatic tracking system in automatic theodolite. The main aspects are following:Firstly, exploit the real-time capture system of video images. Analyze several software proposals of the video capture , in order to reuse the most of programs, and to reduce the difficulty of exploitation,the subject adopted an approach of the video capture program based on VFW. Achieve real-time image capture in VC6.0 platform, and make image can be real-time processed by using the callback function.Secondly, based on analysis of characteristics of the tracking goal and the background, the whole system is divided into four states: the state of searching and capturing, tracking state, memory tracking state and the state of human intervention. Design the automatic detection and recognition algorithm for each state. The paper mainly design and experiment the detection and recognition algorithm in the state of searching and capturing and in the tracking state. In the state of searching and capturing, combine background-difference algorithm and determining the compactness of subjects to detect and recognize the target. In the tracking state, use gray template matching to detect and recognize the target, furthermore, design "heart-shaped region template" for the close tracking and "the whole target template" for the long-distance tracking to match. In order to improve the real-time of the system, use forecast to limit the scope to be matched in the tracking state.Finally, simulation experiment the designed algorithm in the VC platform, analyze and compare the experimental results. The results show that in the state of searching and capturing, the designed algorithm can detect and recognize the target which has a certain compactness, the algorithm is feasible and effective. In tracking state, detection and recognition algorithm are feasible in close or far tracking state. The time of processing an image is about 0.9s, which satisfies the requirement of the subject. The whole algorithms satisfy the expected results of the subject.