Saying the host site was serving malware to users, Google has removed a controversial photo of First Lady Michelle Obama from Google Image Search. The site itself, however, remains listed in Google web search results without any visible malware warning.

Welcome to the murky world of free speech, politics, and Google.

It began last week, when Search Engine Roundtable pointed out a racist image showing as the number one result in Google Image Search for the term [Michelle Obama]. The image was apparently removed yesterday.

In a Google Web Search Help Forum thread discussing the image, a Google employee named Jem explained yesterday that there are three reasons why Google would remove content from its index:

“… Google views the integrity of our search results as an extremely important priority. Accordingly, we do not remove a page from our search results, or images from our Google Images results, simply because the content is in very poor taste or because we receive complaints concerning it. We will, however, remove pages from our results if we believe the image, page (or its site) violates our Webmaster Guidelines, if we believe we are required to do so by law, or at the request of the webmaster who is responsible for the image.“

(emphasis mine)

An offensive photo of Michelle Obama doesn’t obviously violate any of those three guidelines on its own. Google’s press office has yet to respond to our request for an official statement. But in the comments of today’s Search Engine Roundtable post, Google’s Matt Cutts says the site was violating Google’s webmaster guidelines:

“… that page did violate our webmaster guidelines because it was serving malware to users, which violates the quality guideline that says ‘Don’t create pages with malicious behavior, such as phishing or installing viruses, trojans, or other badware.’ I believe that the Images team did a general anti-malware sweep.”

Sure enough, a [site:] search for the site in Google Images produces no results. But the malware sweep apparently didn’t reach the main web search index. As Michael Gray points out on SER, the site itself is still listed in Google.com search results with no malware warning.

Making matters slightly murkier is that, as you see above, the image was hosted on Google’s own blogging platform.

Google’s critics will no doubt call this favoritism toward the Obama administration and be quick to point out the company’s ties to Washington, DC. Google CEO Eric Schmidt endorsed Obama for president, later campaigned with him, and then turned down an offer to join the administration. The Wall Street Journal has reported that Google was the fourth-largest corporate contributor to Obama’s presidential campaign. And we’ve reported about a few notable Google employees who’ve left to work for the Obama administration.

On the other hand, we’ve also written more than once about potential friction between Google and the Obama administration. Christine Varney, the Assistant Attorney General for Antitrust has been quoted as saying that Google “has acquired a monopoly in internet online advertising.” See our stories Will Obama Be The Downfall Of Google? and Google’s Anti-Trust Problem Appears Very Real for more.

If, in fact, the blog hosting the offensive image of Michelle Obama also hosts malware, Google’s removal of the image seems justified in light of the company’s stated policies. But, in that case, a malware warning should also be placed on the site itself in Google’s main search results. Until that happens — and perhaps even after — Google’s critics are likely to question the decision to remove this image.

While today’s episode may be a case of Google looking for an excuse to remove an image from the index, it should be said that the same exact image of the First Lady can be found on other sites and remains in Google Image Search because those sites apparently don’t meet the criteria for content removal. There are similarly offensive images of the President himself that can be found quite easily in Google Image Search, too.

The newest addition to Google Labs is a mashup of Google Images, Google Similar Images and the visual query refinement tool, the Google Wonder Wheel. Called Google Image Swirl, the new tool clusters similar images based on characteristics found in both the text surrounding images, as well as using some of the visual analysis techniques that I wrote about in Google Adds Search By Similarity To Image Search back in April.

Image Swirl results look a lot like the thumbnails presented in regular image search results, but you can see that there are many images stacked behind the topmost result.

Click on a thumbnail, and the stack expands to show you a cluster of images, related by visual characteristics. Click on one of those, and you’ll get a further expansion, this time based on textual or conceptual relationships.

Image Swirl currently works for about 200,000 of the most popular queries. You can tell which queries have Image Swirl results when you start typing in the search box, as available searches will auto-complete, similar to Google Suggest. Google said it plans to expand the number of Image Swirl results available in the future.

Six months ago, Google released an image search labs project that allowed you to search by similarity within image search. Google announced that they have promoted that labs project as a default feature in Google Image Search.

Now when you search on Google Images, you may often find a link under the image that says, “Find similar images.” When you click that link, Google will show you images that look similar to the image you were looking at.

Here is a picture of how that link looks:

I personally tried it this morning, but for some reason, non of the example searches work, nor do my own examples work. It is possible that this feature is not fully launched yet or that Google has pulled it due to a temporary bug. I am not sure, but if you do not see it, you can always play with it at similar-images.googlelabs.com.

Google Operating System spotted two new Google search options including custom date ranges web search and advanced image searches directly on the left side navigation.

When Google first launched search refinements in the “search options” section, they did not allow you to add a custom date range. Now you can, here is a picture:

It also seems like Google is testing the “search options” feature for image search. I personally do not see it on my computers, but Google Operating System took a screen shot:

Postscript: Later on today, Google officially announced the addition of search options on Google Images.

If you go to Google Images and search on a keyword, and then click on new SafeSearch filter drop down (might only work on a PC), you will see a new way of reporting offensive images. A drop down lets you select, “report offensive images.” Here is a picture:

When you click it, the page changes, to give you check boxes and a way to report several images as offensive. Here is a picture of that page:

This does help people report several images at once. It does not seem to work for me on any Mac browser.

Google Operating System, like always, spots that Google has added the creative commons filter to the advanced image search page. In June, Google added the filter through advanced URL parameters, but now it is officially added to the advanced search page.

Here is a screen capture of the filter options, but you can learn more on usage rights here.

Yahoo added the same type of filter back in May.

Postscript: Google has now posted about this new image search feature.

Google Image Search has made it a bit easier to toggle between the three levels of SafeSearch. SafeSearch is the adult image filter Google uses to help prevent adult content from showing up in the image search results. The new toggle switch allows you to click on the filter type and then select from a contextual drop down menu to make the filter moderate, strict or turn it off. Here is a picture:

I believe the first to write about this was Google Blogoscoped, yesterday at about 3:30 pm (EST).

When most of us travel, we like to visit famous landmarks in the area we’re touring. It’s easy to find famous landmarks, like the Eiffel Tower or the Great Pyramid, but can be difficult to find lesser known sites. And for armchair travelers, it can take time to search for images of landmarks even with powerful image search tools that use GPS tagged photos or offer virtual tours.

A group of researchers from Google and the National University of Singapore have helped solve this problem, by building a “web-scale landmark recognition engine.” They’re presenting a report on their work today at the Computer Vision and Pattern Recognition (CVPR) conference in Miami, Florida.

This is similar to the work UCSD researchers did using Google data centers a couple of years ago. That work, called Supervised Multiclass Labeling (SML), involved systematically training a computer to recognize statistically similar objects, and tagging the images with labels that described image characteristics. For more on that project, see Teaching Google To See Images.

By contrast, the landmark recognition engine first had to gather the names of landmarks, and then associate those names with photos of specific landmarks. The researchers used clustering techniques to find similar images, filter out “noisy” images, as well as eliminating false positives (images that look similar but depict two entirely different things).

The researchers compiled a comprehensive list of landmarks from more than 20 million GPS-tagged photos and images located on online tour guide web pages. Armed with this list, candidate images for each landmark were obtained from photo sharing websites and by querying an image search engine. The resulting landmark recognition engine successfully identifed images of more than 50,000 landmarks from all over the world with 80% accuracy.

Unfortunately, Google has no immediate plans to make the new landmark search engine public, or incorporate it into image search. But you can read more about it in this official Google blog post, A new landmark in computer vision, or if you really want to dive into the technology, read the paper the researchers are presenting: Tour the World: building a web-scale landmark recognition engine (PDF).

A couple weeks after Yahoo Images added the creative commons search filter, Google quietly added their own to Google Image Search.

The option is not available in the user interface, but you can set the filter by adding a parameter to the URL. For example:

• Public Domain Images: Add &as_rights=cc_publicdomain to the URL
• Creative Commons Attribution Images: Add &as_rights=cc_attribute to the URL
• Creative Commons Attribution Share-Alike Images: Add &as_rights=cc_sharealike to the URL

Google Operating System created a special search box with radio buttons that allows you to search for these types of creative commons filters without adding the parameter to the URL.

Blogstorm has noticed that Google is showing several images within the UK search results for a search on personal development books.

This is unique in that there are 9 images under a search result’s title and above the description. Here is a screen capture, furnished by Blogstorm:

In addition, it shows “17 images” in the snippet information, which indicates that Google understands that there are indeed a total of 17 images. The images on that landing page do seem significant to the rest of the page. I imagine that Google is thinking this page is more about image search than web search. I also do not think this has anything to do with the new rich snippet support because I don’t see the microformat code on the page and rich snippets currently only work in the US, not in Google UK.