Colour-based lips segmentation method using artificial neural networks

A colour-based technique for lip segmentation is presented throughout this work. Basically, we make use of colour spaces to categorize pixels as either lip or non-lip using artificial neural networks. This study clearly shows how a novel method for fusion of the existing colour spaces practically produces better results than individual colour spaces. More accuracy in dealing with face images under many different conditions has been achieved. Comparing this work with other researchers´ work using the same databases, we found that our method, which involves the fusion of different colour information that comes from different colour models, outperforms other methods.