The search for digital images is evolving
For Digital Future | September 29, 2008
The images may contain large amounts of information but are difficult to categorize by the search engines of the network. Various groups of researchers are investigating ways of cataloging the images of the network and will allow search by content, not just by name or description.
The images contain much information, much more than in hundreds or thousands of pages of text. But the information contained in an image is purely visual and therefore does not tell us anything about when or where the picture was taken, what people in the room or what objects we can see. This makes cataloging and image search is particularly difficult.
Scientists at a major European research project called aceMedia have taken significant steps toward a solution to this problem. They are building a layer of information in digital image files. His vision is that the image files contain information divided into two parts: the metadata (information for use on the Internet) and an intelligent layer that automatically generates word of the image data.
An extra layer of information, that information manually and automatically added to the images are generated when you can revolutionize the search for images on the Internet and on computers or mobile phones. The technologies developed by aceMedia have attracted interest from companies who want to take advantage of these developments to exploit the investigation in different directions.
One of the technologies used by aceMedia is software that can identify visual low-level descriptors, such as areas of color that can be sky, sea, sand, snow, and information about the texture, edges and shapes.
Combining the descriptors of low-level contextual sets of rules, such as domain ontologies (such as the fact that the blue zones at the top of an image is likely to be heaven, or the beach and snow is unlikely appearing on the same image) convert these data into a rich source of information.
Convert these low descriptors into useful information is very difficult, but the team is aceMedia cumin low-level data with the results of advanced sensors, which are commercially available in some cameras today. All these processes add more tracks and data to be able to search for images.
Another layer of information can be added by users. They can add rules that define their personal preferences, profiles and policies to create a system of personal registration.
AceMedia has already carried out a framework that defines the API and thus standardize the development and export it in a multitude of applications.
At about 5 five years, a good number of these technologies will be in common use. It also will be combined with other tools such as geo-labels used in GPS receivers.
The more distant future may bring us anything, surely the cameras will detect the place where it has been done to define the image and thus a large number of useful parameters for searches. This, combined with a powerful search engine that is able to interpret visual information from an image file can be a revolution
For more information visit the Web aceMedia.
Topics: Photography, Technology |
- 10,000 million images on Facebook
- Jpg image files to hide
- CUIL, the new Internet search engine
- Visual search engine, SearchMan
- All the media files from NASA available at NASA Images