摘要
在交通行业日益发展的今天,车牌识别技术对于公路车辆监管以及车辆轨迹跟踪越来越重要,考虑到庞大的车辆信息数据量,单机处理数据能力已不能满足实时性的要求。本文在详细研究分布式处理平台Hadoop的工作原理后,利用其强大的HDFS存储系统与MapReduce数据处理方案,通过Java对Matlab的调用,简化了识别程序,搭建了分布式处理平台,即使在数据量庞大的情况下也能够进行车牌识别分布式计算。实验结果表明,在处理2 000张以上的车牌图像时,运行效率提升了2倍左右。
With the development of transportation industry,the technology of license plate recognition plays a more important role in the traffic regulation and vehicle tracing than before. Present bulkiness of vehicle information and the capacity of stand-alone data processing can not meet the real-time requirements. The powerful storage system of HDFS and data solution of Map Reduce are used to improve license plate recognition algorithm after studying the principle of Hadoop( a platform of distributed data processing). By Java of Matlab calls,the recognition procedure is simplified,and a distributed processing platform is set up. The license plate recognition can be distributed in computing with a large amount of data. Experimental results show that when dealing with more than 2 000 images of license plates,operating efficiency improved about 2 times.
出处
《桂林理工大学学报》
CAS
北大核心
2016年第2期383-387,共5页
Journal of Guilin University of Technology
基金
广西自然科学基金项目(2013GXNSFAA019334)
关键词
云计算
车牌识别
混合编程
分布式计算
cloud computing
license plate recognition
hybrid programming
distributed computing