In the Spring of 2014, I worked on a real-time skin detection algorithm with a graduate student. Our motivation for this project was to implement skin detection to develop a robust, fast hand pose estimation system and develop gesture recognition applications on Google Glass. The algorithm is based on random forest classification using color model features and Laplacian Maps to help discriminate skin pixels from background pixels and noise pixels. I implemented the system using OpenCV 2.4.7.
Click the link here to view research paper, pubilcation pending.