Towards robust skin colour detection and tracking

Skin colour modelling has been a popular research topic over the years due its possible practical applications. Varying skin tones makes modelling of skin colour as one of the most complex and challenging task. Accuracy and performance of skin detection and tracking highly depends on the skin model in use and the tracking method. In order to perform in real time, skin detection and tracking algorithm has to be computationally efficient and robust. This paper presents our effort on designing such an algorithm with a number of simple cascaded filters. It has been successfully implemented and tested with number of popular independent image databases available on the internet. This algorithm guarantees an accuracy rate of 70%.