How Do Image Search Engines Work?
Other People Are Reading
-
Definition
-
Image search engines are online programs designed to locate specific images available on the Internet. Overture, Northern Light, Google Image Search, Ithaki Image and Photo Metasearch, IXQUICK, Lycos Multimedia Search, FAST Multimedia Search, Ditto, Picsearch, Yahoo! Picture Gallery and Picsearch are examples of the image search engines. Search engines are programmed to search in three different ways: by the manner in which the graphic file has been saved, by keyword or by the titles listed on the web page.
Search by Graphic Files
-
Graphic files may be saved using a variety of methods, but the most commonly used methods include .gif (Graphic Image File), .tif (Tagged Image File) and .jpg (Joint Photographic Experts Group). Video is usually saved as an .avi (video) file. Image search engines using this search method locate any file listed using the saved extension. For example, a programmer creating a site includes several images saved as .jpg files with the ".jpg" term embedded in the programming language. This search engine combs the programming language for graphic file listings, pulling only the "jpg" files. ImageDupeless (shareware), SurfPics (freeware) and Internet Graphic Hunter (Shareware) are three popular graphic file search programs.
-
Search by Keywords
-
Keyword searches work as traditional search engines, with one general word. Searching under the term "horses" would bring up any image that was saved using that term. The main problem with this search method is the individual programmer's ability to properly list the correct subject. Flickr is a popular image site that allows the searcher to enter keywords that may match the terms the image posters have selected to describe the photographs on the personal sites.
Search by Titles
-
This search method includes the title used on the web page. It generally works well, provided the titles align with what is shown in the image, but some titles do not describe the image but simply add text for an article. Fine art images are frequently indexed by titles since many art works have titles, such as the "Mona Lisa" or "The Scream." Some searchers are more familiar with the name of the work over the name of the artist. Artcyclopedia and Pixsy are examples of search engines that allow the researcher to input text to retrieve images.
Human Indexing
-
Some specialized image search engines employ humans to search through websites and list the individual images on each page. The Library of Congress photographic archive is an example of this type of image search. Individual photographs are given an index number and the image is reproduced in various formats, including two sizes of .tif files. A written description is also provided for each image. LOC photographs are also indexed using keywords from the description and by titles. This multi-indexed process is labor-intensive, and most search engine sites lack the funds to achieve this type of extensive documentation.
-
