Improved Adaboost Face Detection
Songyan Ma, Tiancang Du · 2010
This paper presents a color based on the Adaboost face detection methods. This method has less color test detection time, high detection rate and AdaBoost false detection rate is low algorithm, strong adaptability advantages. First, by the statistical characteristics of facial skin color, the color image segmentation, to be a candidate face region; and then use the trained AdaBoost algorithm classifier cascade to make face region detection of candidates and ultimately get precise positioning of the human face. Through experiments show that the Adaboost face detection method based color segmentation improved the detection rate greatly than the original Adaboost method.