Monte Carlo and the perimeter and moment estimators based on it for two-dimensional scene objects. The described estimators were used to build a criterion for selecting a frame from a video sequence, based on which we can count vehicles moving relative to the camera.When applying vision techniques in Intelligent Transport Systems [1], we often encounter a situation where, using only image information, we must count the number of moving objects in a scene while determining their geometric parameters such as surface area and perimeter. For example, this could be a gate counting and recording the size of moving cars on a road for the purpose of automatic traffic volume estimation or automatic traffic control, along with acquiring statistical data on vehicle sizes for further analysis. Classic methods for estimating the geometric features of two-dimensional scene image objects require significant computational effort and, in many cases, cannot be used in practice for this reason (especially in industrial applications with low computing power). (…)
The article contains 11,770 characters.
Source: Logistyka Magazine
A fast method for counting vehicles based on camera images – PDF file for download