基于云计算平台的图像识别技术研究  

Research on image recognition technology based on cloud computing platform

作　　者：康文龙[1] 王广桢 蒲志新[1] 丛佩超[1] KANG Wen-long;WANG Guang-zhen;PU Zhi-xin;CONG Pei-chao(School of Mechanical Engineering, Liaoning Technical University, Fuxin 123000, China)

机构地区：[1]辽宁工程技术大学机械工程学院,辽宁阜新123000

出　　处：《传感器与微系统》2018年第6期30-32,35,共4页Transducer and Microsystem Technologies

摘　　要：应用图像处理技术、支持向量机(SVM)算法以及Hadoop处理技术,将云平台与SVM算法结合起来,研究基于云计算平台的图像识别技术。通过交通标志识别实例验证方法可行性与识别效率。研究结果表明:基于所提技术的识别精度要高于基于反向传播(BP)和云计算平台的图像识别技术以及单机运行的基于SVM分类器的图像识别技术。在节点数达到一定数量后,识别效率高于单机平台运行的图像识别技术。Use image processing technology,support vector machine（ SVM） algorithm and Hadoop processing technology,combine cloud platform with SVM algorithm,to study image recognition technology based on cloud computing platform. The feasibility and recognition efficiency of this method are verified by traffic sign recognition.The research results show that the recognition precision of image recognition technology of cloud computing platform is higher than that of image recognition technology based on back propagation（ BP） and cloud computing platform and single operation image recognition technology based on SVM classifier. When the number of nodes reaches a certain number,recognition efficiency is higher than that of single platform running.

关 键 词：云计算 HADOOP平台 图像识别 支持向量机 交通标志 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]