Researches On Moving Target Detection And Tracking For Images Of Shooting Ranges

Optical measurement equipments are among the firstly and most widely used measurement equipments in shooting ranges. The ballistic trajectory and gesture parameters of the flight object, which are the significant evidences for the assessment of weapon test and the analysis of failure, can be obtained by interpreting the images captured by the optical measurement equipments. The interpretation of the images captured by the optical measurement equipments is taken as the background of this dissertation. The automatic detection, tracking and positioning of the moving target, measurement of 3 dimensional gesture parameters, matching of multiple targets and other relative technologies are investigated in this dissertation, and these technologies are applied successfully to the interpretation system of the images captured by the optical measurement equipments. The main contents and achievements of this dissertation are as follows:1. The function of the optical measurement applied in the measurement and control of shooting ranges is presented, and the development of the image interpretation equipments and technologies is reviewed. The main problems of the image interpretation are summarized as: automatic and reliable detection of moving target, accurate tracking and positioning of target, measurement of 3 dimensional gesture parameters, and matching of mutiplue targets.2. The electro-optic theodolites and high speed photography cameras which are most frequently used in shooting ranges are taken as examples to illustrate the structure and principle of the optical measurement equipments, and the processing requirements and procedure of the images. The imaging model of the the optical measurement equipments is set up, and the imaging relation is deduced, which establishes the foundation for the following discussion of the technologies such as target detection, tracking and positioning, etc.3. Moving target detection is the first step of image interpretation, and it was done manually before. With the requirements for the efficiency of interpretation and realtime target detection and tracking, it is necessary to investigate the moving target detection algorithm. Image differencing is a widely used method in moving target detection, but as most of the work only used the absolute value of difference map and the motion of the background must be compensated, the robustness and practicality of the method have been reduced greatly. A new moving target detection method which uses both the positive and negative difference maps is proposed in this paper. The new method accomplishes moving target detection according to the associations of the target's gray, shape and motion in both positive and negative difference maps. The robustness and practicality of the method has been proven by experimental results. 4. The tracking and positioning of targets in the long series of images were done manually in traditional interpretation, as a result the efficiency was low, and the operators had great burdens, and worst of all, the interpretation accuracy droped with the time increased, so an accurate target tracking algorithm is necessary which can adapt to the rotation and zoom of target. The Least Squares Image Matching(LSIM) method which was used in photogrammetry is introduced to moving target tracking, and a Weighted Least Squares Image Matching(WLSIM) based tracking algorithm is proposed. The WLSIM based algorithm sets weights according to the target's shape for the Least-Squares Image Matching Algorithm, as a result matching error produced by the background in the tracking window can be avoided. Experimental results are shown to demonstrate the robustness, efficiency and accuracy of the proposed algorithm.5. It is difficult for traditional interpretation method to obtain the 3 dimensional gesture parameters, while the application of digital image analysis technology in image interpretation gives a new chance to overcome such problem. A conclusion was drawed by analyzing the methods to measure the 3 dimensional gesture parameters, which is the key of measurement of the 3 dimensional gesture parameters for the common target of shooting ranges is the automatic and accurate extraction of its line contour. A new boundary detection algorithm which makes use of both the magnitude and direction of gradient is presented. The algorithm first calculates the histogram of gradient's direction, which is defined as the sum of the magnitude of gradient of the pixels whose directions of gradient are similar. The main or wanted direction of edge can be easily distinguished based on the histogram. Then the positions of the boundaries are found by searching through the whole image according to the main direction. Finally the moment method is used to obtain the precise positions of the boundaries. Unlike traditional boundary detection algorithms, which use only the strength of the edge and need complex operations, the new algorithm requires no threshold and works automatically. The proposed approach is tested by a variety of images with encouraging results.6. It is a difficult problem to distinguish a large amount of similar small targets in the processing of images. According to the motion relevancy of image sequence of single station and the epipolar geometry restriction of targets of multi-stations, a new technology for interpretation of multi-targets of multi-stations is proposed, which makes use of Kalman filter and epipolar geometry, and breaks through the restriction of single satation image interpretation. And a method for calculating the error strip of epipolar lines which can restrict the taget matching is proposed. The correctness and feasibility of the method have been proven by simulation data.7. Image interpretation system is an important equipment for optical measurement post-processing. A design scheme for universal interpretation system is proposed based upon the analysis of requirements for interpretation of films, videos and digital images. Several key technologies which were not discussed in the previous chapters are introduced, i.e., synchronization of image capture and film step, mathematical morphology based extraction of time mark on high speed photography images, recognition and extraction for dots matrix code, feature based character recognition, the calibration and error correction of the film digitizing camera and the state transform of detection and tracking.A universal interpretation system, which is competent for the interpretation of films, videos and digital images, is realized by integrating those technologies discussed above and making innovation. It has the advantage of high accuracy and automation and is easy to use. The system has many applications in several shooting ranges, and the version of its software has been upgraded continually according to the user's suggestions which made the system more perfect and practical. The system has been playing an important role in the interpretation of images of shooting ranges.The target detection and tracking technologies proposed in the dissertation can be applied to the realtime detection and tracking of moving targets in shooting ranges, and some other areas such as image guidance, intelligent transportation, etc.