基于Spark的分布式车流量检测方法设计与实现

Design and Implementation of Distributed Traffic Flow Detection Method Based on Spark

针对现有车流量检测算法处理批量视频效率较低的问题,提出了基于Spark分布式计算框架的车流量检测方法;为提高计算节点对待处理数据的读取速率,在底层采用HDFS将交通视频元数据进行高效存储,同时设计了分布式帧差法以实现车流量并行检测与统计;最终,检测结果由主控节点存储于HBase中,以提升数据访问的可靠度;在真实数据集上的测试结果表明,与传统帧差法实现车流量检测算法相比较,处理速度提升528%,同时准确率为90.5%。

Aiming at the problem that the existing traffic flow detection algorithm is dealing with the low efficiency of batch video, a traffic flow detection method based on Spark distributed computing framework is proposed. In order to improve the reading rate of the data to be processed by the computing node, HDFS is used to store the traffic video metadata efficiently, and the distributed frame difference method is designed to realize the traffic detection and statistics. Finally, the test results are stored in the HBase by the master node to improve the access reliability of the data. The results of the test on the real data set show that compared with the traditional frame difference method to a- chieve traffic flow detection algorithm, the speed increase 528%, while the accuracy rate of 90. 5%.