| As a research hotspot of computer vision, the technology of moving target detecting and tracking on video is being researched widely, meanwhile a prominent result has been achieved. Thanks to the emergence of open source library, the efficiency of video project development has been improved greatly.This dissertation studies and designs a multi-sensor target tracking system, based on OpenCV. This system applies to moving target detecting and tracking in specific scene; the accuracy of tracking and position is improved by collaborative working of multi-sensor.Firstly, we analyze the development situation in tracking field, then discuss the on demand and general design of the system. Secondly, this dissertation gives the detailed design of the core modules of the system, the processs of target tracking, the process of target position measure, the process of parallel processing, and the target fusion on multi-sensor. Finally, we describe the details of the realization.In the experiments parts, the accuracy of target tracking, as well as the position measure, in specific scene, are proved by a big amount of experiments. In addition, the quality of time consuming has been taken consideration via some other experiments. |
