Auto-recognition Pedestrians Research Based on HOG Feature and SVM Classifier for Vehicle Images

In order to estimate the potential hazards and adopt strategies for preventing the accidents, this paper is about the research of auto-recognition pedestrians. Using the captured images of the vehicle recorder, based on the HOG feature processing technology and SVM classifier, the automatic detection and recognition of pedestrians are realized. In the past, the cost of pedestrian identification system is too high and its universality is poor. In this research, optimization methods of reasonable sample allocation and difficult training are used to make the training model more general and effective.