Research on Vehicle Re-recognition Algorithm of Image Feature Fusion Based on YOLOv3
In order to improve the accuracy of vehicle re-recognition,a method combining global and local features is proposed,which solves the problem of low license plate recognition accuracy caused by factors such as blurred license plates,unclear vehicle contours,and occlusion during vehicle re-recognition.Firstly,this paper uses Siamese Network to match the rear shape,front appearance,and overall vehicle shape of the vehicle images to be detected.Then,it uses the LPRNet based on YOLOv3 to combine the vehicle local shape,overall shape,and license plate recognition for vehicle re-recognition.The results show that the proposed method can achieve vehicle re-recognition under the changeable road environment,and the comprehensive accuracy of re-recognition reaches 93.63%,which is 7.28%,3.08%and 0.75%higher than the re-recognition model of DRDL,OIFE and RAM,respectively.
国家科技期刊平台
NSTL
万方数据
