Scene text reading based cloud compliance access: User behavior analyze of graphical system and images data sanitization

Cloud Compliance Access system assure that cloud users could utilize resources of cloud platform while conforming to regulations and the specific functions of this system were User Behavior Analyze and Data Sanitization. However, the problem of information extraction from images data is still to be addressed for both of these functions. The authors, in this paper, ventures to propose a new cloud compliance access system based on scene text reading, which could realize timely auditing to graphical systems of users’ operations and image data sanitization on users’ transparency. To improve the feasibility and availability of this system, with regard to cloud computing scene, the author refined the model of scene text reading and realize two main functions of cloud compliance access through extracted information through this model. In evaluation part, the author examines the accuracy rate of scene text reading model through real data (the overall accuracy rate achieved 93.4%), simulated real application of this mode as well as evaluate the effect of two main functions, user behavior analyze and data sanitization.