Human face detection based on genetic algorithm
Zhang Jun-chang, Yi Zhang · 2010
To overcome feature redundancy in the construction of human face detector with AdaBoost algorithm, an improved human face detection method is proposed. First, eight new rectangle feature types are proposed and AdaBoost algorithm is used as a feature selector to make rough selections. Then genetic algorithm with strong search ability is introduced to optimize those selected features and their parameters to build a system that can search out most human faces in images with lower false positive rate and less number of weaker classifiers. Simulations show that compared with existing AdaBoost algorithms, the proposed method can effectively remove feature redundancy, reduce false alarm rate and achieve a higher detection speed with much more accuracy.