基于融合算法的车牌识别系统

License Plate Recognition System Based on Fusion Algorithm

摘要: 普通的车牌识别系统一般是使用单一算法对车辆的车牌号码进行识别，识别结果往往会受到很多外在因素的影响，从而降低识别准确度。针对于这一现状，本文提出一套基于融合算法的智能车辆车牌识别系统，通过基于Sobel边缘检测算法来进行车牌的定位与提取，截取出符合系统要求大小的车牌图片，再通过SVM分类算法进行对比，选出合格的车牌图片。选出后的车牌图片利用人工神经网络算法来进行车牌字符的识别，得到完整的车牌信息。经过实验证明，多种算法的融合作用实现的车牌识别系统整体有着高速且精准的识别性能。

Abstract: Ordinary license plate recognition systems generally use a single algorithm to recognize vehicle license plate numbers, and recognition results are often affected by many external factors, thus reducing recognition accuracy. In view of this situation, this paper proposes a set of intelligent vehicle license plate recognition system based on the fusion algorithm. Through the Sobel edge detection algorithm to locate and extract the license plates, intercept license plate images that meet system requirements, and then compare and select qualified license plate images through the SVM classification algorithm. The selected license plate image uses the artificial neural network algorithm to recognize license plate characters and get complete license plate information. The experiment proves that the license plate recognition system realized by the fusion of multiple algorithms has a high-speed and accurate recognition performance.