Pure vision-based feature post-fusion vehicle re-identification method

According to the vehicle re-identification method based on pure vision feature fusion, vehicle fine-grained features and a global feature extraction model structure are improved, and a new measurement method is provided for vehicle orientation constraint. Firstly, a classification model is designed for vehicle type information of a vehicle, the classification model comprises a UNet structure used for extracting fine-grained features and a convolution model used for image coding, and a classification result of the model is used for sample pre-screening. And for the global features of the vehicle, coding is carried out by using a convolutional neural network and a model of a Transform structure, and post-fusion of the features is carried out. Meanwhile, a local center loss function based on a clustering algorithm is introduced to constrain the intra-class distance of the targets in the same direction, and finally the sum of the cross entropy loss, the triple loss and the local center loss is used for representi