Abstract
In this paper, the SIFT feature, HOG feature and SVM are adopted to design a movement vehicle recognition system, when the vehicle is relatively close to the camera and it backs to the camera, also the vehicle can be implemented to track. In this paper, two feature extraction methods, SIFT feature and HOG feature are respectively used to extract the features from 6000 images of vehicles with back facing the cameras and 6000 images without vehicles. Then SVM is used to train them to obtain a model. Then, we use the moving object recognition algorithm to extract the moving object in the given video, and put each moving object into the trained model to judge. If the model is judged as a vehicle, it will be tracked.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have