摘要
本文利用Enomaly虚拟云架构技术设计了一个云手写识别系统,除了提供高准确率的识别功能外,同时使得倾斜书写识别、用户自适应识别等需要高计算及存储资源的功能实现成为可能。实验结果表明,传统C/S模式的传统服务器在用户并发数为300时处理能力已经达到极限,而采用基于云计算架构的手写识别系统能轻松处理1000个并发用户的服务请求,在处理300个并发用户时,接入率为100%,远高于传统服务器模式的接入率(82.7%),平均识别处理时间仅为16ms,大大低于传统服务器模式的处理时间(340ms)。
We design and implement a cloud handwriting recognition system based on enomaly virtualization cloud technology. In addition to providing efficient, high accurate recognition services, our system can support orientation free and user adaptive service, which need high computing and storage resources. Our experiment show that, using traditional C/S model, the server would reached its capacity limit when 300 concurrent users are requesting service. While using cloud-based architecture, our handwriting recognition system can easily handle 1000 concurrent users＇ request. In dealing with 300 concurrent users, the access rate was 100%, much higher than the traditional C/S model of the access rate （82.7%）, the average recognition processing time is only 16 ms, significantly lower than the traditional C/S model of the processing time （340 ms）.
出处
《电信科学》
北大核心
2010年第9期84-89,共6页
Telecommunications Science
