The Design And Implementation Of Content Management System Based On CBIR

For enterprises,with its continuous development,it will produce a variety of content resources,such as documents,web pages,picture material and so on.To improve the reuse efficiency and retrieval efficiency of these resources will have a great positive impact on the efficiency of the staff.In order to solve the above requirements,the content management system came into being.Traditional content management systems have been able to address most of the needs of enterprise content management.But its retrieval is often based on the text,with the increasing importance of picture content today,it has some limitations.According to the above,this paper takes the Baidu Inc internal project RMP,a content management platform based on CBIR(Content-Based Image Retrieval)as the background,and makes a research on the following content:Firstly,research,design and implement a content management system based on business requirements.System function covers the life cycle of contents,from the creation of content to the release of content.To study the solution of content management related upload,download,version control and so on.Secondly,research on content based image retrieval(CBIR).Research on the existing image retrieval algorithms,such as the color feature based image retrieval algorithm,contour feature based image retrieval algorithm,SIFT algorithm and so on.Finally,we choose the Vector of Locally Aggregated Descriptors technology which is suitable for this system,and the implementation and related technologies are studied and analyzed.Lastly,CBIR technology is integrated into the system,in the image content upload,and image retrieval when the application of CBIR technology,to provide users with an additional choice of image retrieval.To test the accuracy of image retrieval system.At present,the system has been put into formal use,used in the development of a number of Baidu Inc,and in the further promotion.