基于肤色模型的人脸检测算法

在此研究了彩色图像中的人脸检测算法,在肤色模型方面讨论了肤色在不同的色彩空间中的表现,通过对一定数量的样本建立了一个图像库,选择色彩空间并建立了肤色模型,然后通过形态学等处理后得到候选区域。然后对候选区域使用特征脸的方法进行验证。通过这样的方法在保证了精度的同时克服了特征脸方法原有的搜索速度慢的特点。最后本文对所使用的方法作了一个总结并对下一步的工作进行了展望。

The paper works over the algorithm of face detection in color image.Firstly we discuss the skin＇s Performance in different chrominance space in the discussion of skin color model.During the Statistics of some sample skin to build a imagelibrary,we can build the skin color model and using the morphological process select the candidate region.Then the candidate region can be validated by the eigenface method.Using this method we can not only ensure the accuracy of the detection but get over the defect of eigenface method＇s slow search as a pyramid.In the end of the paper,bring forward a summary about how to expand the detection system and outlook the follow-up work.