复杂地铁人流环境下多视角模糊人像鉴定技术

由于监控拍摄到的人像整体模糊、分辨率低,不易鉴定辨别,因此提出了复杂地铁人流环境下多视角模糊人像鉴定技术。根据整体视觉系统与经典三色信号之间的存在性对数关系,确定模糊图像的三色适度取值,获得有效的图像光照补偿处理,其次通过进项去噪和分辨率重置剔除图像中冗余信息,来保证还原图像的精准度,随后使用基于简单时态网络(Simple Temporal Network,STN)的三维人脸矫正技术,矫正人像姿态以及角度,最后将经过处理的图像输入至支持向量机LIBSVM中,建立人像模型的底层数学模型,通过目标函数值完成鉴定。仿真结果证明,所提方法可以有效对模糊人像进行矫正鉴定,且具有精准度高、过程简单的优点。

Multi-view Fuzzy Portrait Identification Technology in Complex Metro Flow Environment

In this article,a multi-view identification method for fuzzy human image in complex subway streams environment was proposed.According to the logarithm relationship between the whole visual system and the classical trichromatic signal,the trichromatic fitness value of fuzzy image was determined,and the effective image illumination compensation was obtained.Moreover,the redundant information in image was eliminated by input denoising and resolution reset,so that the accuracy of the restored image was guaranteed.And then,3D face correction technology based on STN(Simple Temporal Network)was used to correct the pose and angle of human image.Finally,the processed image was input into the support vector machine LIBSVM,and the bottom mathematical model was built.Thus,the identification was completed by objective function values.Simulation results show that the proposed method can effectively correct and identify the fuzzy human image.Meanwhile,this method has the advantages of high accuracy and simple process.