文件名称:A-FaceDetection-Based-on-Frontal-InverseAdaboostAl
介绍说明--下载内容均来自于网络,请自行研究使用
人脸的检测和跟踪对于自由立体显示非常重要。本文采用连续Adaboost算法学习出一种瀑布型的人脸检测器,同时
针对adaboost算法误检率比较高的弊病,提出了一种正反Adaboost检测的思想,最后通过利用已有积分图以及最小值滤波,完成
了对人眼的识别操作。实验结果表明,算法对现有adaboost算法精度上的改进令人满意,基本能满足自由立体显示的需要。
-Accuratelylocatingandtrackingthepositionoftheusefsfaceisimop rtantforautostereoscopicdisplay.A
casacdefacedetectorlearntbyusingraelAdabo stalgorithm ispresented,andforthepurop seofreducingthefalse
alarm,anovelprocedurecalledFrontal—InverseAdabo stisproposed,thenaprocesstogainaccurateeyes’loac—
tionsisexecutedwithminimum filterna dintegralimage.A setofexperimentsarepresented.theresultsindiacte
thattheprecisionimprovingtonormalAdabo stalgorithm issatisfying.Itacnbeseenthattheproposedmethodis
veryefifcientandhaslotsofsigniifcantvaluestoautostereoscopicdisplay.
针对adaboost算法误检率比较高的弊病,提出了一种正反Adaboost检测的思想,最后通过利用已有积分图以及最小值滤波,完成
了对人眼的识别操作。实验结果表明,算法对现有adaboost算法精度上的改进令人满意,基本能满足自由立体显示的需要。
-Accuratelylocatingandtrackingthepositionoftheusefsfaceisimop rtantforautostereoscopicdisplay.A
casacdefacedetectorlearntbyusingraelAdabo stalgorithm ispresented,andforthepurop seofreducingthefalse
alarm,anovelprocedurecalledFrontal—InverseAdabo stisproposed,thenaprocesstogainaccurateeyes’loac—
tionsisexecutedwithminimum filterna dintegralimage.A setofexperimentsarepresented.theresultsindiacte
thattheprecisionimprovingtonormalAdabo stalgorithm issatisfying.Itacnbeseenthattheproposedmethodis
veryefifcientandhaslotsofsigniifcantvaluestoautostereoscopicdisplay.
(系统自动生成,下载前可以参看下载内容)
下载文件列表
一种基于正反ADABOOST的人脸检测算法.PDF