In this procedure, a novel colour space modelling is introduced that improves face skin extraction performance by using multiple thresholds as decision criteria. A three-step technique is also used to reduce noise and other negative impacts. The first critical stage in automated facial image analysis is frequently face identification, which tries to identify the existence along with subsequent location of human faces in an image. For the purpose of enhancing the effectiveness of extracting the colour of the face's skin, this procedure adopts a novel colour space modelling that employs several thresholds as decision criteria. Additionally, a three-step approach is employed to reduce noise and other negative impacts. The results of the experiments demonstrate that the suggested method could efficiently extract the side face contour line and should offer a theoretical foundation for the real-time tracking of fatigue applications.