一种适用于井下人脸跟踪的改进主动形状模型

摘要

For low accuracy of face tracking applying active shape model(ASM) in condition of sudden change of illumination and contaminated and covered miner face in underground coal mine,an improved ASM was proposed.Firstly,mirror sample set is formed by definition of mirror image with selected special training sample set.Then,the mirror sample set is processed by logarithmic scale.Finally,related block model is used as ASM learning model for training.The experimental results show that the method can effectively improve accuracy of underground face tracking.%针对煤矿井下光照变化较大、矿工脸部受污染及遮挡等情况下主动形状模型(ASM)应用于井下人脸跟踪精度低的问题,提出了一种改进ASM.首先选用专用的训练样本集,通过定义镜像图像形成镜像样本集,然后对镜像样本集进行对数尺度化处理,并用相关块模型作为ASM的学习模型进行训练.实验结果表明,改进ASM能有效提高煤矿井下人脸跟踪精度.