基于监督局部映射的煤矿井下人员身份鉴别方法

为提高煤矿井下人员身份识别率,在局部保持投影(LPP)算法的基础上,提出监督局部映射(SLP)算法。该方法充分利用数据的局部和非局部信息及类别信息,对数据进行维数约简,使特征空间同类数据间的距离更小,不同类数据间的距离更大。该方法能够克服煤矿井下艰苦、空间受限环境中人脸、虹膜和指纹识别率不高的问题。在真实步态数据库上的实验结果表明,基于步态的煤矿井下人员身份鉴别是可行的。

Personnel Identification Based on Supervised Locality Projections in Coal Mine Underground

To improve the recognition rate of the personnel identification in coal mine underground,based on the LPP,a SLP algorithm was worked out and applied to the personnel identification in coal mine underground.SLP makes full use of the local information,non-local information and class label information of the data for dimensionality reduction of the original data.After dimensionality reduction,the distance between the data points in the same class becomes smaller in feature space,while the distance between the data points in the different classes becomes larger in feature space.SLP can overcome the problem that the recognition rates of face,iris and fingerprint in underground coal mine are not higher than that in the regular environment.The experimental results on the real gait databases show that the personnel identification in coal mine underground based on gait is effective and feasible.