基于统计模型的人体携带危险物检测

摘要

To detect the concealed objects in active millimeter-wave images,the human statistical models based on generalized compound probability density function and generalized K probability density function were established referring the parametric model for the human millimeter-wave images.A method of concealed objects detection was presented based on body structure,statistical model and constant false alarm rate.The experiments of statistical model were conducted.Results show that the fitting degree reaches 99％ for millimeter-wave images using generalized compound probability density function and generalized K probability density function.And the experiments of concealed objects on images of actual acquisition were conducted.Results show that the accuracy rate reaches 93.51％ and the false alarm rate reaches 9.82％,the effectiveness and practicality of the proposed method are verified.%为研究人体携带危险物的检测,针对人体毫米波图像,采用参量模型的建模方法,建立基于广义复合分布和广义K分布的统计模型;在此基础上,基于人体结构的先验知识,利用人体统计模型,采用恒虚警率对毫米波图像中人体携带危险物进行检测.对实际获取的图像进行仿真实验,实验结果表明,广义复合分布和广义K分布与毫米波人体图像的拟合程度达99 ％,危险物的检测率为93.51％,虚警率为9.82％,验证了所提人体图像统计模型和危险物检测方法的正确性和有效性.