Computer vision is a broad field with millions on use cases in every day life, widely used in robotics, image processing and video analysis. In my thesis I propose to implement an algorithm to track moving objects on a recording with using the open source software library of OpenCV. The chosen method is the popular approach of extracting and matching feature points. These points, that are easy to recognize on the picture, and their position can be localized are easy to follow from one frame to another.
The real life problem is to use the application in a vehicle counting system, thus to extract feature points of moving cars. These information can be used in traffic engineering, to plan parking places and effects of roadblocks.
The implementation of a framework is necessary to test and compare various algorithms of feature tracking. I also developed a program that gathered the steps of achieving the purpose of the task with the ability of the easy extension.
The thesis summarizes briefly the theory behind the used technology and even gives a short introduction to OpenCV.