基于自适应背景的目标跟踪算法研究

Research on target tracking algorithm based on adaptive background

摘要:

随着现在的社会发展以及经济进步,我国的科学技术方面发展迅速,特别是在技术监控方面更是突飞猛进。为了更好的对目标遮挡影响进行降低,我国在这方面主要依据自适应的技术发展背景下提出目标跟踪计算法,用来完善我国的监督控制技术。这种计算方式第一是根据对观察目标的基本外观形态进行的鉴定与跟踪,将其自身的运动量进行平均计算；其次是根据时空的运行方向与特征进行跟踪目标的计算,建立比较完善整体的运行模型,再根据这个运动模型以及整体的状态对监督目标进行检测与控制,这期间就会形成一种遮挡掩膜。对于掩膜是一种将程序数据等绘制成光刻板,在程序使用期间非常可靠,并且制造成本比较低,使用方便；最后是在不同的使用情况下将不同参数进行收集,自动的适应运动模型的运行。针对这种计算方式的实验主要是利用两种在国际上经常使用的CAVIAR、York数据进行测试,并且根据这两种数据对测试的精准度与多重目标跟踪等进行评定,检测跟踪的整体性能。通过多方面的研究表明这种方式的跟踪的性能非常好,并且还能很好的将跟踪目标的鲁棒性进行遮挡。

Abstract:

With the social development and economic progress, China's rapid development of science and technology, especially in terms of technical monitoring is a rapid development. In order to reduce the impact of the target occlusion, the target tracking algorithm is proposed based on the adaptive technology development in this area, which is used to improve the supervision and control technology of our country. This calculation is based on the identification and tracking of the first appearance of the basic observation of targets, will exercise its own for the average; second is calculated according to the running direction of the space target tracking and characteristics, to establish a complete model of the whole operation, according to the motion model and the whole state detection and control of the supervision of the target, this period will form an occlusion mask. The mask is a program data mapping into light rigid, very reliable in the program during use, and the manufacturing cost is relatively low, easy to use; the last is used in different situations with different parameters were collected, to adapt to the motion model of automatic operation. According to this calculation experiment is the use of two kinds of CAVIAR, York data is often used in the international test, and according to these two kinds of data to test the accuracy and multiple target tracking evaluation, the overall performance of detection and tracking. Through a number of research shows that this way of tracking performance is very good, and it can be very good to keep track of the robustness of the target.