SBIR Phase I: New Algorithms for PTZ Camera Based Object Tracking

这个小型企业创新研究(SBIR)第一阶段研究项目将调查和评估一类用于视频监控系统中的PTZ(平移-倾斜-变焦)摄像机的新型运动目标跟踪算法。在当今的大多数视频监控系统中，实时目标跟踪是由操作员使用PTZ摄像机手动执行的。这往往压力大、效率低(操作员一次只能控制一个PTZ摄像头)，并导致结果不一致。这项拟议的项目将研究一类新的算法，以引导PTZ相机自动跟踪感兴趣的对象。这是通过使用带有新对象和观察模型的最佳过滤器来实现的。该项目的成果将是可以集成到标准视频监控系统中的软件模块，以提高其能力。视频监控系统是打击犯罪和恐怖主义的重要工具。今天市场上的大多数系统都是相对标准的DVR(数字录像机)，几乎没有智能功能。拟议的创新(自动对象跟踪)是一种智能功能，可以通过允许标准系统获得更好、更有用的图像来显著提高其能力。由于许多最终用户都需要此功能，因此它对设备供应商和集成商具有极大的吸引力。因此，它具有商业潜力。最后，通过引入新的目标跟踪模型(在项目描述中详细介绍)，拟议的创新还促进了图像处理和计算机视觉研究的最先进水平。

This Small Business Innovation Research (SBIR) Phase I research project will investigate and evaluate a new class of moving object-tracking algorithms for PTZ (pan-tilt-zoom) cameras in video surveillance systems. In most of today's video surveillance systems, real-time object tracking is performed manually by human operators using PTZ cameras. This is often stressful and inefficient (an operator can only control one PTZ camera at a time) and causes inconsistent results. The proposed project will investigate a new class of algorithms to direct a PTZ camera to track an object of interest automatically. This is done by using an optimal filter with new object and observation models. The project outcome will be software modules that can be integrated into standard video surveillance systems to improve their capabilities.Video surveillance systems are important tools in the fight against crime and terrorism. Most of the systems on the market today are relatively standard DVR's (digital video recorders) with few smart features. The proposed innovation (automatic object tracking) is a smart feature that can significantly improve a standard system's capabilities by allowing it to get better and more useful images. Since this feature is demanded by many end-users, it is highly attractive to equipment vendors and integrators. Hence, it has commercial potential. Finally, by introducing new models for object tracking (detailed in the Project Description), the proposed innovation also advances the state-of-the-art in image processing and computer vision research.