An Improvement to Media Discovery Service Using Image Identification Technique

Yen Chieh Fu, Lan Yang

The media discovery service is an important element in managing and discovering media objects based on the metadata in Digital Object Architecture. However, types of metadata used today often lack some of the basic requirements for compatibility, such as standard definitions and unique identification. The media discovery service searches for media objects based on the metadata information. Therefore, it is not able to discover those media data with incomplete or missing metadata information. — In this research we propose to the media discovery service for identifying images with missing or incomplete metadata information. The image identification technique by integrating the image identification technique into the service will find the information of identical images from databases when the service discovers any media object without complete metadata information. In the image identification technique the SURF (Speeded Up Robust Features) algorithm is used to extract the interest points and feature descriptions for image comparison and then produce the similarity. In this way the media discovery service may recognize an image without metadata information associated with it. In general, the media discovery service treats two identical images with different metadata as two distinct files. In the improved media discovery service we verify the metadata first before the system registers the homological data into the storage system. This improvement adds to the media discovery service the ability to search for unknown media objects and the advantage to save storage space from avoiding the homological media objects registered into it. Thus, it makes the media discovery service more powerful and more space efficient.

Published
2011-10
Content type
Original Research
Keywords
Media discovery service, reverse image search, image processing, surf, image identification, image metadata, image database
DOI
10.5594/M001068
ISBN
978-1-61482-940-9