利用人脸检测结果，结合混合高斯模型，以人脸肤色为样本在线建立了具有针对性的人体肤色模型。
Since a human face has the similar skin color distribution with the human body, we can use the skin pixels in a detected face to build the skin color model of human body dynamically by GMM.
该方法以有限个理论计算数据为样本，采用多项式函数离线进行回归，在保证高的逼近精度的前提下，以显著提高在线计算速度为目的。
The presented method is more accurate for using polynomial regression of theoretical calculated data as sample off-line and is quicker for the reason of its fixed polynomial pattern on-line.
提出了关键技术， 包括：挖掘主题的定义方法、海量训练样本的在线生成和高性能数据挖掘算法。
The key technologies is proposed, including methods of definition of mining topics, online acquirement of extra large amount of training samples, and algorithms of data mining with high performance.
应用推荐