机器人系统中人脸特征提取技术的研究与实现.pdfVIP

23 06 2006 06 : 1006 - 9348( 2006) 06 - 0176- 04 1 1 2 3 陶宇权 , 程德福 , 李成荣 , 徐喆 ( 1. , 130026; 2. , 100080; 3. , 300160) : , , Adaboost, ; , , ; , , , , , , , , , : ; ; ; : TP317. 4 : B Research andRealizationofFacialFeatureExtraction Technology inRobotSystem 1 1 2 3 TAO Y u - qua , CHENG De- fu , L I Che g - ro g, XU Zhe ( 1. I stitute of E lectric Scie ce a d E gi eeri g of Jili U ivers ity, Jili Jili 130026, Ch i a; 2. I stitu te ofAu tom atio Chi ese A cadem y of Scie ces, Beiji g 100080, Chi a; 3. I stitute of Commu icatio a d I form atio E g i eeri g of T ia ji I dustry U iversity, T ia ji 300160, Ch i a) ABSTRACT:The research a d realizatio of facial feature extractio tech ology i i tellige t robot system are described i th is paper. A ew m ethod of facial feature extractio used i i tellige t robot system is proposed. First, it w ill detect face i origi al im age w ith a algorithm s based o Adaboost, the face mi age is acquired. Seco dly, the acqu ired face mi agew ill pass a space filter, the resu lt im agew ill ot i clude o facial regio a d itw ill pass the process of th resho ld i g. The thresholded mi age w ill be volum ed by a recta gle temp late, a wt o d mi e sio al curve is acqu ired. I the acqu ired curve, the wt o h ighest peaks rep rese t the eye a d eyebrow, the eye, eyebrow a d m outh are acqu ired accord i g to facial feature geom etry distribu ti g relatio u ltmi ately. The proposed m ethod ha