Researches On Auto Target Detection, Tracking And 3-D Pose Measurement For Imagery Of Shooting Range

Optical measurement technique is an important component of measurement and control technology in shooting ranges. The ballistic trajectory and pose 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 automatic interpretation of the images captured by the optical measurement equipments is taken as the background of this dissertation. The automatic target detection, tracking and three Dimensional(3-D) pose measurement are investigated and applied successfully to the interpretation system of the images, which improve the reliability and efficiency of image interpretation. The main contents and contribution of this dissertation are as follows:1. As the information of targets in shooting range is mostly known, two robust methods based on single example training for target detection are proposed as follows:①Target detection method based on steering kernel morphological filter of single example, which makes full use of the stability of steering kernel description and is suitable for target detection in complex background.②Target detection method for similar multi-target detection based on shape entropy difference, the shape entropy difference operator is designed by specific shape of target, which can be used to detect small targets in multi-target images of shooting range.2. In order to track the target stably, three methods for stable tracking are proposed, they are as follows:①A new small target tracking method based on the gradient Laplacian of Gaussian(LoG) operator, which has the high precision of target detection and is robust to the change of illumination.②A drift correction method based on feature points matching, which can be combined with many tracking methods to address the problem of drift. The tracking result is checked frame by frame and drift correction is applied if necessary.③A new method based on point description for multi-target tracking with multi-view constraint. The constraint of multi-view is utilized to analyze the trajectories of targets in image sequence, and the algorithm is very suitable for tracking multi-targets in shooting range.3. To extract the axis of small target in image of shooting range, an iterative moment method for extracting target’s axis is proposed, the iteration process is designed to improve the fit region of target, which can restrain the interference of background obviously, therefore the high-precision result of extraction is obtained. In addition, the influencing factor and error of the axes method is analyzed quantificationally.4. By using the imaging model and target’s structure, two methods for pose estimation are proposed. They are as follows:①New convenient method for measuring rolling angle by axis and off-axis lines of targets, which can be applied to determine the rolling angle of targets such as plane and cruise missile.②New method for measuring the pose of target with known structure based on proportions of feature points, which can determine the 3-D pose from mono-view. Under the parallel projection imaging model, the pose of object can be directly calculated without target’s 3-D position and camera parameter such as focus length.5. Considering the requirements of digital image interpretation, a design scheme for universal digital image interpretation system is proposed. Meanwhile, several key technologies are discussed for constructing the image interpretation system, i.e., infrared image enhancement techniques, analysis of targets’interpretation position from multi-sensor images, parallel interpreting images from different cameras, applications of target detection, tracking and 3-D pose measurement in image interpretation system.The technologies researched and other related technologies for image interpretation are utilized to realize a digital image interpretation system, and the system has been playing an important role in images interpretation of shooting ranges of Navy, Air Force and Gerneral Armament Department.