He Dongjian

Video-Based Detection of Abnormal Behavior in the Examination Room

Aiming at the problems of partial occlusion and background clutter in the examination room, we propose a method for behavior detection using spatial-temporal shape and flow correlation. The method first extracted training templates using interactive video cutout technique, and automatically segmented the video into 3D spatial-temporal volumes using improved Mean Shift algorithm. Then we slide the template across the video and compute the matching distance. We complement our shape-based features with flow, and efficiently match the volumetric representation of an action against over-segmented spatial-temporal video volumes. Thresholding the correlation distance and finding the peaks give us locations of potential matches. The experiment results indicate that this method achieves human’s action detection robustly in crowded, dynamic environment.

Research and implementation on access control of management-type SaaS

In the paper, we analyze the features of access control of management-type SaaS. Based on the traditional RBAC, we put forward the access control model based on both tenant and role, in which the tenant is as the minimum unit of administrative domain. To be sure user identity with physical security, we put forward the hierarchical authentication and management of user in the management-type SaaS. In order to ensure the access control model of management-type SaaS in line with the reality, we abolish the inheritance right of role in the traditional RBAC. Based on the timing diagram of UML, analyzing the access control model of the cattle public administration platform based on the SaaS, we present its dynamic modeling of access control. Test and analysis is shown that the access control model based on both tenant and role can ensure the accessibility, security and privacy to access cross-domain in the management-type SaaS, and promote the popularization and application of the management-type SaaS.

Security Scheme for Sensitive Data in Management-Type SaaS

In this paper, we proposed the potential security problem of sensitive data in management-type SaaS which have no valid evidence when service provider embezzling or responsibility confirmation in service level agreement dispute, and proposed a scheme to ensure sensitive data safety by taking advantage of encryption and signature technique. Non-credible PKG of identity-based signature mechanism is used as signature mechanism for sensitive data in the scheme. Based on non-credible PKG of identity-based signature, a double-layer PKG framework is built by using the sub-PKG of service provider and tenants according to characteristic of the security responsibility of sensitive data is supervised by both service provider and tenants which have ownership and use right of the software respectively. The symmetric cryptography which ensures confidentiality of sensitive data is employed based on balance efficiency and safety. The related questions are also discussed. Security and the feasibility analysis show that this scheme ensures the confidentiality, integrity and non-repudiation of sensitive data. It can also provide valid evidence for service level agreement dispute and impel the application and the development of management-type SaaS. The proposed scheme can be widely applied to interactive distributed system of sensitive data.

A model for image categorisation based on a biological visual mechanism

Abstract For integrating a visual attention mechanism and object recognition in the visual cortex we propose a novel biologically‐motivated computational model for image categorisation. We first extract the focus of attention using an image‐driven, bottom‐up attention model and then adjust it according to the principles of whole effect and centre preference. After that, we obtain the region of interest, depending on the characteristics of object spatial proximity and object similarity. Based on this we compute a set of position‐ and scale‐invariant C2 features and finally pool them into the standard classifier to achieve image categorisation. We test our model on an image database used in SIMPLIcity. The results suggest that our model can not only classify images effectively under various complex “clutters” but also that it needs only a few training samples.