Google Polishes its Image

Google has taken the wraps off the beta version of a new specialized image search capability, allowing you to search and browse more than 150 million digital images (GIF and JPEG files). Keeping tradition with its flagship text-based search engine, Google's image search results are generally highly relevant, even for basic one or two word queries.

Image search is a major challenge for search engines, since they're designed primarily to handle text. Most web images don't contain text, so the search engine is forced to rely on other clues to figure out what an image represents. The most basic clues are the filename and text surrounding the image on the page, which may -- or may not -- describe it.

Beyond that, engines can read the contents of the image ALT tag, if a webmaster has used it, which displays an alternate description of the image for browsers that can't display images. Finally, links to pages with images often act like text captions, offering a summary of what the image is about.

"Google image search looks at dozens of factors to determine the image subject matter to ensure relevant images are returned to the user, leveraging similar ranking technology used for Google's HTML and PDF search results," according to Google spokesperson David Krane.

To use Google image search, enter your search terms in the image search box. The result page displays 20 images at a time to facilitate fast browsing of image thumbnails. Most other image search engines only show 10 images/page.

The following information is returned next to each image:

- Image thumbnail
- Height and width of the image
- URL of the image framed by the referring URL
- Image size (in kilobytes)
- File name

Duplicates are removed and buttons, icons, and banner ads are filtered from the search results.

Clicking on a thumbnail brings up a framed display with a slightly larger version of the thumbnail (or the entire image, if it's small enough) in the top half of the frame, and the page the image was found on in the bottom half. From the top frame, you can click the thumbnail to display the full-size image, remove the frame to display the entire page, or return to your search results.

Displaying search results in this way this offers helpful context, but the use of horizontal frames is somewhat awkward and wastes space. Vertical frames, similar to the search result display within Internet Explorer, would provide just enough room to display the thumbnail on the left side and show much more of the source page on the right side of the screen, especially for users with larger monitors. But this is just a minor quibble.

All of Google's advanced search commands are available with image search. You can also limit your search to a specific kind of image file by using the "filetype:" operator in your query. For example, if you wish to see images of balloons that are in .jpg format, then enter [ balloon filetype:jpg ” in the search box.

Google's image search works particularly well when there are many images available on the web for Google to choose from. A query for "Miles Davis," for example, returned 1,830 results. The results included an impressive mix of images, ranging from album cover art to photos of the jazz trumpeter.

The few false drops generally occurred when there was a lot of relevant text on the page in a table format, and the (incorrect) image Google associated with the text was in the row immediately above or below the text. This isn't a huge problem, since the page still contained relevant results and the (correct) image could usually be found by examining the page.

Image search offers an intriguing way of discovering new web sites that might not turn up in standard text results. A search for "mars," for example, turned up expected results from JPL and the NASA image archives. But it also produced images from other sites that turned out to be excellent astronomy resources that I was unfamiliar with.

In fact, a quick comparison of the results of text vs. image searches for "mars" showed that there was little overlap, and yet both sets had high-quality, relevant links. So you might consider doing an image search as an alternate way of querying Google for an entirely different "view" of your search terms.

Google Image Search

Google Image Search FAQ

AltaVista Launches Paid Inclusion Program

AltaVista has quietly introduced Express Inclusion, a new program that lets webmasters submit up to 500 URLs to the global database that powers AltaVista's search results. With Express Inclusion, AltaVista joins Inktomi in what appears to be a major trend toward charging admission to crawler-based search engine databases.

Danny Sullivan will take a closer look at AltaVista Express Inclusion in an upcoming issue of Search Engine Report, describing its features and assessing its competitive threat to the other major search engines. To ensure you get his review and analysis, make sure you are signed up for the Search Engine Report.

AltaVista Express Inclusion

Search Headlines

NOTE: Article links often change. In case of a bad link, use the publication's search facility, which most have, and search for the headline.

About the author

Chris Sherman is a frequent contributor to several information industry journals. He's written several books, including The McGraw-Hill CD ROM Handbook and The Invisible Web: Uncovering Information Sources Search Engines Can't See, co-authored with Gary Price. Chris has written about search and search engines since 1994, when he developed online searching tutorials for several clients. From 1998 to 2001, he was's Web Search Guide.