Back-Propagation 알고리즘 과 SVM을 이용한 차량 번호판 인식 구현에 관한 연구

A Study on the Implementation of the Vehicle Licence Plate Recognition System by Back-Propagation Algorithm and SVM

This study will suggest methods for a license plate recognition system that is suitable for license plate identification, separation of letters, and recognition of letters in order to recognize a licence plate efficiently. A method with fast real-time processing that has no omission of license plate territory is needed. License plate territory must be extracted stably when input image is obtained at night even with imbalance of light. In order to do this, uniformity of image was pursued through proper preprocess and a method with no omission of license plate territory was used for image that includes error. The suggested algorithm can overcome drawbacks of back propagation by using SVM. However, as SVM can create many sorters, back propagation was used to find the best solution that SVM had selected.