자동차 번호인식률 향상을 위한 전후 처리 알고리즘 구현

Abstract

Recently, as the use of car becomes more common, License Plate Recognition(LPR) that can prove the uniqueness of a vehicle are becoming more necessary. LPR can be used for various purposes such as parking control, parking guidance information using camera, illegal parking control, speeding, etc. Therefore, accurate identification of the car number is essential. Therefore, researches for increasing the recognition rate using various image processing and pattern recognition technologies are continuing. In this paper, we propose how to improve the recognition rate by pre- and post-processing the car images received from the camera. In the pre-processing part, it is necessary to apply variable thresholds filtering in parallel to provide an optimal environment for increasing recognition rate. In the post-processing part, it is necessary to apply voting method for increasing recognition rate. This system is able to obtain a higher recognition rate than another LPR system even in a number of factors (light, place, and circulation) that may cause degradation of recognition rate using proposed pre- and post-processing methods.