In order to reduce computational complexity ,a fast filtering algorithm with EyeMap infor-mation fusion is proposed . Firstly applying image segmentation based on skin-information ,we ob-tained the interested eyes region that narrowed the scope of the search and reduced the false tolerance rate;Secondly Haar-Like characteristics was used in the of cascade AdaBoost eye detection that getting the underlying eyes area ;Finally acquiring the accurate eye location through the fast algorithm based on EyeMap information ,then then binary-processing ,estimating the closed degree of eyelids based on calculating eye's axial out-connected rectangle . According to the of fatigue PERCLOS judgement standard ,it can analysis the eye of fatigue condition .T he experiments show that the improved algo-rithm can judge quickly and efficiently the eye state of driving's fatigue .%为降低人眼检测过程的计算量,提出一种改进的融合EyeM ap信息的快速过滤算法。利用基于肤色信息的图像分割,得到感兴趣的人脸区域,缩小了搜索范围,减小了检测的错误接受率;然后使用Haar-Like特征进行级联AdaBoost人眼定位检测,得到潜在的人眼区域,使用EyeMap的快速算法获得准确地人眼区域,再进行二值化处理,计算眼睛的轴向外接矩形以估计出眼睑的闭合程度,结合PERCLOS评判标准进行疲劳状态分析。实验证明:改进算法能快速实时有效地识别驾驶员疲劳时的眼部状态。
展开▼