Human-Computer Painting System Using Facial Mouth Feature Information

As computer vision technology been developing, while camera becomes standard configuration for personal computer (PC) and computer calculating speed becomes faster and faster. Computer vision have been more popular used in people daily life, people can use computer as an eye to see and recognize the real world. In this paper, we will present a novel electronic head Painting system which is based on human facial feature. Our system firstly detect human face from the input image by using face features in YUV Color model, and then for face candidate, use the nearly reversed relationship information between U and V cluster of face feature to detect mouth region. Then in the detected mouth region, mouth lip region gray level is usually lower than other region but V level higher than others, so we can reversal mouth region gray level and use associated threshold method to pick up mouth region pixels. Geometrical relationship between mouth region and face region boundary can be used to confirm human head rotation angles correspond to camera, which can be used to control brush movement in screen. Then we extract mouth shape feature by segment it into 5*5 pieces and calculate the eigenvector. Got eigenvector can be inputted into recognition machine based on neural network to recognize the mouth shape. Output codes of recognition machine can be judged as command to control painting brush when and with which size of brush to paint. We have made an experiment on windows XP to evaluate the effect of our system. The experiment result showed that our system can represent a good effect for head painting system.