The Research And Optimization Of Face Detection

Face detection is used to obtain the size and location of a person's face accurately based on the digital computer recognition technology,it highlights the main facial features,and weakens other minor information like clothing and environment.With the deepening of the study,face detection can also reflect more accurately characteristics of the eyes,nose,mouth,etc.In the field of video surveillance,video coding,intelligent nursing,human-computer interaction,image retrieval and biometrics,face detection technology is applied widely and deeply,which has great research value and application prospect.This paper mainly studied the application of face detection in intelligent video surveillance system,analyzed the current development of face detection technology,introduced detection algorithm based on geometric features,statistical theory and knowledge,and analyzed the advantages,disadvantages and difficulties of mainstream detection algorithms,summed up the most effective solution for face detection.The third chapter improved skin color algorithm because of the brightness played an important role in the skin color,the skin color points were hard to be extracted from non skin color points,and it can not produce obvious clustering.YCrCb color space is to separate luminance and color components,thereby ignoring the Y component,which has great help on the effect of face detection,so the color image is transformed from RGB space to YCrCb space,and through removing the candidate region and setting ROI,you can quickly lock the suspected face region,which can improve the efficiency of detection.Lastly in the face detection part,based on the openCV vision library and Adaboost algorithm,the paper extracts Haar feature and obtain cascade categorizer through integral image,then trained it by Adaboost algorithm,realized the initial face detection,and tested under various conditions finally.