Face Recognition Algorithm Research Based On Internet Of Vehicle

Nowadays,face recognition,once a science fiction technology,has quietly entered people’s lives,and has penetrated into more fields.As the product of the concept of “intelligence” in the field of transportation,“Internet of Vehicle” is gradually developing.More and more new technologies and concepts are put into this field.Focusing on the driver’s face recognition in the cab,face detection algorithm,light preprocessing algorithm and face recognition algorithm are studied,and a simulation verification system is developed to verify the availability of the algorithm.In the aspect of face detection,connected domain is used to detect the rough location of the face,and Adaboost algorithm is used to locate the face accurately.By analyzing the principle of connected domain detection,the key parameters of connected domain detection are determined.ORL face database and Pascal VOC database are used to train Adaboost classifier.In the light preprocessing,wavelet algorithm is used to decompose the face image into the low frequency subgraph containing illumination information and the high frequency subgraph containing face details.And then two kinds of subgraphs are dealt with in different ways.The effectiveness of the preprocessing algorithm is verified by the effect image.In face recognition,LBP,HOG and Broad Learning System are used to extract face features and classify them.The type of LBP Operator,sampling radius and the number of sampling points are determined by comparing the LBP feature map of human face.Through the analysis,the key parameters of HOG algorithm are determined.The BLS network model is trained by adjusting the number of mapping nodes.The accuracy of the hybrid algorithm is 95.60%,which is higher than that of the single LBP,HOG and BLS algorithm.Through the test of simulation verification system,it is proved that the hybrid algorithm has a good effect,can ensure the successful operation of face recognition function and adapt to the change of light intensity to a certain extent.In the case of nonuniform illumination and “Yin Yang Face” to a certain extent,it is able to ensure the normal operation of face recognition function.And it can adapt to the posture change of head to a certain extent.