煤矿井下人员人脸轮廓提取方法

Facial contour extraction method of underground personnel

针对煤矿井下人员脸部轮廓难以提取而导致考勤系统识别率低的问题,提出了一种基于主动轮廓模型和主动形状模型的人脸轮廓提取方法。首先由主动形状模型估计初始轮廓位置,然后通过定义主动轮廓模型的能量函数,采用主动轮廓模型算法多次迭代后缩小标记点与真实轮廓之间的差距,提取由主动形状模型获得的正交插值的灰度值轮廓,即可得到人脸轮廓。实验结果表明,该方法较传统主动轮廓模型能够更精确地提取人脸轮廓形状。

For low identification ratio of attendance system of coal mine caused by difficult facial contour extraction of underground personnel, a facial contour extraction method was proposed based on active contour model （Snake model） and active shape model （ASM）. Firstly, the initial contour position is estimated by ASM. Secondly, the gap between marker and real contour is narrowed after multiple iterations by Snake algorithm through defining energy function of Snake model. Finally, gray value contour of orthogonal interpolation obtained by ASM are extracted, so as to get facial contour. Experimental results show that the method can extract facial contour more accurately than traditional Snake model.