Implementation of Real-Time Skin Segmentation Based on K-Means Clustering Method

Localization and detection of body parts (in this case, hands) is an exigent issue in image processing, since it is a prerequisite for applications like hand gesture recognition. Through this paper, a proposal for implementing an efficient way for segmenting the skin tone is developed by applying appropriate Computer Vision techniques and K-means Clustering on each frame captured by the camera under different illumination conditions as well as in the complex backgrounds, is developed. In this method, detection of skin portions using color identification, and segmentation from the image is done for its implementation in real-time systems. While the paper focuses on using skin segmentation for the detection of hand movements, the application of this approach can easily be implemented in various applications involving Human Computer Interaction (HCI). Examples of such include mouse cursor movement, media player application, writing text on electrical documents, controlling robot, detects the pointing location, sign language, hand posture and face recognition.