Design And Implementation Of Image Contrast System Based On Public Cloud Platform

Purpose of this article is to design and implement an image comparison system by using public cloud service as infrastructure.The practical background of the system is to providing image comparison to a multinational pharmaceutical enterprise.This organization has serval R&D centers within the globe,very large amount research images would be uploaded every day from each site.Medical images are large and complex,it would take huge amount of resources to do the image comparison manually.Therefore,real-time scaling is very important to the infrastructure of the application,public cloud which provides highly scalable compute resource and almost infinity storage is highly suitable for such kind of application.This article did a detail comparison of the largest public cloud providers in the market(Amazon,Microsoft and Google),researched the services which would be used for implementing the system and the prices of them and made a decision of the most suitable platform.The infrastructure of this system is implemented on Amazon AWS,by using the Virtual machine,database,object storage,message queue services.Cloud platform provides very well stability,scalability and economic efficiency.The software infrastrucure of this system including gateway,scheduler and worker software.Those software are implemented on AWS virtual machines,images and the data which is required of image comparison are stored in object storage and database on cloud.The comparison software is coding under boot spring frame by using JAVA,which lower the complexity of the coding.Also AWS which is chose as the cloud platform has a good supportive to the JAVA,meanwhile it provides a JAVA container service called Elastic Beanstalk which makes the implementation easier.This system using open source software OPENCV to do to comparison job,OPENCV can load serval kinds of algorithms from its library to compare images.Different algorithms provide different comparison accuracy,this article listed and compared serval common used algorithms and chose the suitable one based on the application requirements.The application is running stably and normally after it was implemented on AWS platform,never found system crashing or any other circumstance.But the data amount is small for the testing purpose,it still needs to be monitored when under large pressure.