目的 随着军事侦察任务设备的发展，红外与可见光侦察技术成为军事装备中的主要侦察手段。研究视觉目标跟踪技术对提高任务设备的全天候目标侦察、目标跟踪、目标定位等战场情报获取能力具有重要意义。目前，对视觉目标跟踪技术的研究越来越深入，目标跟踪的方法和种类也越来越丰富。本文对目前应用较为广泛的四种视觉目标跟踪方法进行研究综述，为后续国内外研究者对目标跟踪相关理论及发展研究工作提供基础。方法 通过对视觉目标跟踪技术难点问题进行分析，根据目标跟踪方法建模方式的不同，将视觉目标跟踪方法分为生成式模型方法与判别式模型方法。分别对生成式模型跟踪算法中的均值漂移目标跟踪方法和粒子滤波目标跟踪方法，判别式模型跟踪算法中的相关滤波目标跟踪方法和深度学习目标跟踪方法进行研究。首先分别对四种跟踪算法的基本原理进行介绍，然后针对四种跟踪算法基本原理的不足和对应目标跟踪中的难点问题进行分析，最后针对目标跟踪的难点问题，给出对应算法的主流改进方案。结果 针对视觉目标跟踪相关技术研究进展，结合无人机侦察任务需求，对跟踪算法实际应用中存在的重点解决问题与相关目标跟踪的难点问题进行分析，给出目前的解决方案与不足，探讨研究未来无人机目标侦察跟踪技术的发展方向。结论 视觉目标跟踪技术已经取得了显著的进展，在侦察任务中的应用越来越广泛。但目标跟踪技术仍然是非常具有挑战性的问题，目标跟踪中的相关理论有待进一步完善和改进，由于实际应用中的场景复杂，目标跟踪的难点问题的挑战性更大，因此容易导致跟踪效果不佳。针对不同的应用环境，结合具体不同军事装备的特点，研究相对精确和鲁棒并且满足实时性要求的视觉目标跟踪算法，对提升装备的全天候侦察目标信息获取能力具有重要意义。
Objective With the development of military reconnaissance mission equipment, infrared and visible light target reconnaissance techniques have already become the main means of reconnaissance in military equipment. Carrying out researches on infrared and visible light object tracking technology is of great significance to improve the capabilities about battlefield intelligence acquisition and precision strike in military mission equipment, such as: all-weather target reconnaissance, object tracking, and target location. At present, with the rise of computer vision technology, visual object tracking technology has gradually become the focus and difficulty of research, and the methods and kinds of object tracking techniques are more and more abundant. In this paper, four kinds of visual object tracking methods, which are widely used at present, are reviewed. It provides a basis for the follow-up research work on the theory and development of object tracking. Method By analyzing the difficult problems of infrared and visible object tracking technology, according to the different modeling methods of object tracking, the visual object tracking method is divided into generative model method and discriminative model method, the mean shift object tracking method and particle filter object tracking method in generative model tracking algorithm, the methods of correlation filtering and deep learning object tracking in discriminative model tracking algorithms are reviewed in this paper respectively. Firstly, the basic principles of three standard object tracking algorithms, including mean shift object tracking method, particle filter object tracking method and correlation filters for object tracking method, are comprehensively analyzed. Then, listing the shortcomings of the basic principles of the three tracking algorithms respectively, and giving its corresponding difficulties in object tracking that need to be solved, through the analysis of the difficult problems in object tracking, the mainstream improvement scheme of the corresponding object tracking algorithm is given. According to the characteristics of infrared image and the difficult problem of infrared object tracking, the improved algorithm of infrared correlation filter for object tracking is presented. In this paper, we studied the methods of object tracking in deep learning, and divided them into two categories: one is to take neural network as the target feature extraction method. Regarding the mainstream object tracking method as tracking method for tracking framework, analyzing the principle and characteristics of neural network feature extraction, the feature extraction strategy of neural network in object tracking algorithm is given. Besides, the corresponding improvement scheme is also provided according to the characteristics of infrared object tracking. Another one is the object tracking method based on neural network framework, in accordance with the development of the neural network framework summarizes the research on object tracking method, the principle and characteristics of each neural network framework are introduced, analysis of various neural network architecture advantages and shortcomings in object tracking, and to solve the problem of infrared object tracking, combining with the characteristics of infrared image and difficult problems of infrared object tracking, an improvement scheme about neural network infrared object tracking are proposed. Finally, we summarize the current situation of object tracking technology and discuss the practical application and future development trend of object tracking technology. Result At present, the visual object tracking technology has good performance under short-term object tracking condition. However, in the situation of long-time tracking required in practical application is more difficult due to the application scene is more complex, the difficult problem of object tracking is more prominent. Especially the key and difficult problems in object tracking, such as target occlusion and target out of view, so the robustness and precision of object tracking technology are required to be higher in practical application, and needed to put forward corresponding solutions to the problem of long-time object tracking. In view of the research progress of related technology of visual object tracking, combining with the demand of unmanned aerial vehicle reconnaissance mission and the characteristics of unmanned aerial vehicle high maneuverability, this paper analyzes the key problems that need to be solved in the practical application of object tracking algorithm and the difficult problems related object tracking, gives the current solutions and shortcomings which are still existing in the program, and also explores the key issues and development direction of the object tracking technology in the future unmanned aerial vehicle reconnaissance mission. Conclusion Up to now, the visual object tracking technology has made remarkable progress. The accuracy and success rate of object tracking have been significantly improved. It is more and more widely used in reconnaissance missions of military equipment. However, the technology of object tracking is still the very challenging issue. The related theories of object tracking need to be further perfected and improved, especially in view of the characteristics of infrared object tracking. In order to improve the object tracking effect in infrared image, the corresponding object tracking method and improved scheme need to be further studied. Because of the more complex scene in practical application, the difficult problem of object tracking is more challenging. Therefore, it is necessary to require the robustness and accuracy of the object tracking algorithm higher than before, otherwise easy to lead to the phenomenon of object tracking failure. In addition, the object tracking algorithm has high requirements on real-time performance. Therefore, on the one hand, the speed of object tracking needs to meet the accuracy and robustness of different reconnaissance tasks, and on the other hand, improving the tracking speed is the need of real-time requirements. According to the different application scenarios, combining with the application characteristics and the scope of application of different military equipment, it is of great importance and significance to study the visual object tracking algorithm which is relatively accurate, robust and meets the real-time requirements to enhance the equipment"s all-weather reconnaissance ability and target battlefield information acquisition capability.