Face Detection Technology Research Based on AdaBoost Algorithm and Haar Features

摘要

This chapter described a face detection technology that was based on haar features and AdaBoost algorithm. The principles of face detection were introduced, which are haar features, AdaBoost algorithm, and cascade classifier. In the experiment, two different classifiers were trained and tested on both static pictures and video stream. For static pictures, the detection rate reached to 84.6 % (using Classifier 1 in Picture 3) and 92.3 % (using Classifier 2 in Picture 3). In video stream, faces were detected accurately. Experimental results indicated that face detection using haar features and AdaBoost algorithm can achieve high detection accuracy while a small amount of haar features, namely common features, can reduce hardware resource cost and minimize computation time. It is beneficial for improving detection rate in face detection.