Face detection and tracking is important for auto-stereoscopic display. This learning through Adaboost algorithm in a row a waterfall-face detection, while against the ills of adaboost algorithm than the high rate of false detections, come up with a positive and negative detection of Adaboost, finally filtering through the use of integral graphs as well as the minimum value, of identification of human eye operation was completed. Experimental results show that algorithm for adaboost algorithm accuracy improvement is satisfactory, can meet the needs of auto-stereoscopic displays.




File list:
一种基于正反ADABOOST的人脸检测算法.PDF