Virtual Lines Sensors for Moving Object and Vehicle Counters

摘要

Any method of counting objects and vehicles using computerized systems in a real live environment needs to have reliable results and flexibility of operation. The proposed work describes a method of counting moving objects and vehicles within a live or recorded video stream using multiple virtual line sensors that give flexibility to the users in analyzing any location for the number of vehicles within the given video, as existing traffic surveillance cameras may be sited in many positions/angles. The proposed research is an improved object counter. The results of this method have shown a promising accuracy of close to 100% detection in daylight and 91% at night