Rapid Human-Eye Detection Based on an Integrated Method
Xia Liu, Dong Yongqing, Li Su, Huang Chao · 2010
Considering the defects of traditional methods of human-eye detection, such as time-consuming and poor accuracy, etc, a fast and accurate human-eye-detection algorithm has been proposed. First, locating the facial part in the whole image with the Adaboost algorithm, then, with the feature extraction method locating the probable human-eye area, and finally, finding the accurate eye-location with an improved template-matching method. The proposed algorithm integrates the feature extraction method and the template-matching method by making use of their advantages. The proposed eye-detection algorithm can effectively reduce the eye-detection area, and improve the detection speed and accuracy. As the experimental results show, this method can detect human eyes fast and effectively under the condition of no glasses sheltering.