Combination of Algorithms for Object Detection in Videos on Basis of Background Subtraction and Color Histograms: A Case Study

Theo Gabloffsky, Ralf Salomon


This paper presents a combination of algorithms for an object detection and recognition in videos. These algorithms are based on a background subtraction and an histogram comparison. The algorithm were implemented and used for the detection of curling stones in videos from a dataset. These dataset includes three different types of videos, which reaches from (1) only the curling stone is on the over (2) an athlete is behind the stone and (3) an athlete moves in between the field of view from the camera. While analysing the videos, the time was measured which the algorithms needed for their calculations, As the results show, the implemented algorithms are able to recognise position of the curling stone with an detection rate of 100% under best circumstances and with 71.11% under worst conditions.


Paper Citation