Apply the new technology human face recognizing to the identification authentication in the mine, constructs the human face database suitable to miners. Uses the method of wavelet transformation, uses the Haar characteristic, constructs the cascade sorter based on the weight, between multistage sorter each levels through Delta rule adjustment weight, and carries on the person face recognition. Uses the real miner photograph as the sample training structure sorter, it obtains the classified characteristic document based on the XML in the person face and the human eye examination, the experiment effect is good. Completed the person face and the human eye examination prototype system based on the video frequency and the image.%为了将人脸识别这一较新的技术应用到矿山身份验证过程中,首先构建了符合井下人员特点的人脸数据库.识别方法采用小波变换的方法,利用 Haar特征,构建了基于权重的级联分类器,多级分类器每级之间通过δ规则调节权重,文中采用真实矿工图片作为样本进行训练,构造初始的分类器,将该方法应用于人脸与人眼识别,得到用于人脸与人眼检测的基于XML的分类特征文件,进行人脸、人眼识别,实验效果良好,有较高的人脸检出率.并完成了基于视频流与图像的人脸与人眼检测的原犁系统.
展开▼