Images in Search Engine Results

When a search engine presents the results of a search to you, it will often show you the title of a page, a short snippet from the page or a meta description, and the URL of the page. While that information can help searchers choose which pages to visit, showing images from the pages listed may provide a helpful clue about what those pages are about.

But if a search engine started including pictures from web pages in search results, how would it go about deciding which image to show to searchers?

Choosing Pictures for Search Results

If you publish pages on the web, do you include pictures on those pages? How meaningful are those images by themselves, out of the context of your web page?

How well would they describe the content that might be found on your web page if they were presented alone, or with the title, snippet, and URL that might be shown in a search engine result? If those images were made thumbnail size, how well might they represent your pages? Would you want a search engine to show an image from your page next to a search result for your page?

If you’ve used the Cuil Search Engine, you probably noticed that they do try to include images with search results. You may have also noticed that the images they display aren’t always very helpful in providing information about the pages they are matched with, or may not be the best of choices from the pages they appear upon. Hopefully they are fine tuning their use of pictures.

If pictures in search results are something we will see more commonly in the future, we may need to think more about how the images we use on our pages will appear in places like search engine results.

Google will often show images associated with news articles in Google News, and in Google News results that sometimes appear in web search results, next to the snippet that accompanies the result. Those tend to be pretty relevant for the articles that show up next to.

Microsoft published a patent application today that discusses how they might choose images from web pages to show to searchers in search results. While the processes they use might not be what other search engines would use, the patent filing might help us think about how other search engines might also use pictures.

Dominant Media Elements

Microsoft’s approach not only discusses using pictures in search results, but also how those pictures might be chosen if there is more than one image appearing upon a page. Images are referred to in the document as “media elements” and the use of that phrase had me wondering if they would consider using other types of media in this method, such as videos.

An image that is important, and informative about the page that it appears upon is called a – dominant media element – one that is “substantially related to a topic or theme of its display page.”

A search engine might determine how dominant each image is on a web site in an automated manner – having people determine that would take too many people. But, human reviewers might be used initially to train a program that might then use what it learned in that training to make the determination on its own for other sites.

A dominance score would also be based upon features of the images. An image with the highest score might be considered the most dominant media element on the page, and might be used in search results alongside the title, description, and URL for the page. If a page has multiple dominant elements, the one that is most relevant to the search query might be the one used.

The patent application is:

Adding Dominant Media Elements to Search Results
Invented by Ming Jing Li, Shuming Shi, Wei-Ying Ma, Zhiwei Li (Beijing, CN)
Assigned to Microsoft
US Patent Application 20080313177
Published December 18, 2008
Filed: August 21, 2008

Abstract

A method and system for determining dominance of the media elements of display pages is provided.

The dominance system provides a scoring mechanism for scoring the dominance of media elements of display pages based on features of each media element of the display page.

To generate the scores for the media elements of the display page, the dominance system first identifies the media elements and then identifies the features of the media elements.

The dominance system then scores the identified media elements using the provided scoring mechanism and the identified features.

Using Features to Decide Which Images are Dominant

A page about the White House, in Washington, D.C., might have a photo of the White House and a picture of the Presidential seal. The White House photo might be considered the dominant media element of the page, and the picture of the Presidential seal could be considered to be a non-dominant media element. The decision of which image to show would be based upon looking at different features.

Features that might be looked at could include things such as the size of the image, and its location on the page. In one version of the process described in the patent application, the dominance scoring system might consider three different types of features; image level features, page level features, and web site level features.

Microsoft explored using image level features, page level features, and website level features in a previous patent filing that described how images could be used to help rank web pages.

I wrote about that method in a post titled How Search Engines May Use Images to Rank Web Pages. There are some more details about features of images in that patent application that don’t appear in this one, but this one tells us more about the assumptions behind the features they might consider.

Image level features would be taken from the image, and could include:

Image size – the width of the image times the height of the image (we’re told that dominant images tend to be bigger than non-dominant ones).

Aspect ratio – the minimum of the width and the height divided by the maximum of the width and the height. (again, we are told that dominant images tend to have larger aspect ratios than non-dominant images).

Image quality – Such things as sharpness, contrast, colorfulness, blurness, and so on. Sharpness could be the ratio of “clear” edges to all edges. Contrast might be defined as the ratio of brightness of the foreground to the brightness of the background. Colorfulness would be the number of colors of the image. (Dominant images supposedly tend to be sharp and colorful images with high contrast.)

Image category – whether the image is a photograph and if it contains a human face. (Again, we’re told that dominant images are more likely to be photographs that contain human faces than non-dominant images.)

Page Level Features

The patent makes a few assumptions in the section about image level features. One made for page level features is that the most important and informative images of web pages are often placed in a position where they will attract attention from visitors to the page.

Position of the image on the web page – the position might be identified by looking at the x and y coordinates of the image within the web page. We’re told that dominant images tend to be located at the center of the top of the web page.

Area ratio of the image to the web page – the ratio of the image size to the web page size. Again, dominant images supposedly occupy larger areas of their web pages than non-dominant images.

Website Level Features

There are images that appear on a site in advertisments, logos, as decorations, navigation, bullet points, and so on. While some of these images may have high image qualities, and could be located in important areas of web pages, they may not help a searcher understand what a specific page might be about if seen next to a search result for that page.

We’re told that these “noisy” kinds of images tend to have some common characteristics:

Image is external to the site – advertisements tend to be hosted on other sites.

Image is duplicated on many pages of the site – non-dominant images such as logos tend to appear on many pages of a web site, or a certain percentage of pages.

Conclusion

The possibility of a search engine including images from your web pages in search engine results might get you to think about the quality of the images you’re using on your pages, and how well those images reflect the content of those pages.

If a search engine showed images from your pages in search results, would you be happy with the images that they show?

A couple of interesting articles I found about using images in web pages that stress the importance of using meaningful images that are related to the content of the pages they appear upon:

Share

8 thoughts on “Images in Search Engine Results”

  1. Images are very important aspect in Search Engine Optimization. Images could be used to gain more traffic to your website. Thus, you must your the best images for your contents with good alt description.

  2. Thank you.

    I do see pages where alt text is used for images that doesn’t actually describe the pictures, but is relevant for the terms and topics used on the page, and that’s not really a good approach.

    While images and their associated alt elements, and captions can be helpful to SEO, I think it might be a good idea to focus first upon how those images actually make pages more informative.

    The pictures you put on your pages can be a very helpful aspect of design and communication, providing people with a visual approach to communicating the ideas that you present on your pages. If your pages provide information, images can be used to present that information in a non-textual manner. If your pages describe and sell products, giving people a chance to see those products is essential.

    The pictures that you use should be relevant to the content and focus of your pages. If they are, then any alt attribute descriptions you add to describe those images are going to be relevant to the pages as well, and should also help in the SEO of those pages.

  3. Applying keywords into the Alt text is always a great idea. You must however do it sparingly. You do not want to take a black hat approach. Always only relate to the image.

  4. Image optimization is great for both image search results and seo since alt tags are also counted as keywords. But as Nick Stamoulis mentioned you need to be careful of not overdoing it.

  5. Pingback: » Pandia Weekend Wrap-up December 21
  6. Hi Nick,

    I would limit that even more. Images that are used for decorations or navigation should use empty alt text (alt=”"). Images that are used to add value to the content of a page should ideally be relevant to the content of the page, and the alt text used for those should use keywords as appropriate. Alt text should be an alternative description for the image shown – so use images that related to what the page is about, and it’s more likely that it will be appropriate to use your keywords in the alt text for those images, and in a caption for the image as well.

    Hi SEOSapien,

    Agreed – using an image that can be found in image search can help bring people to your pages, and optimizing that image well can help your page be found in organic web search. I think this patent application points to an even more interesting idea – that the images you choose might be displayed besides the search result for your page – which means that choosing unique, interesting, and engaging images for you pages may be a way in the future to get more visitors to your site.

  7. Most of the website have implemented this technology to make search result more user attractable and colorful. In some cases search engine users may not know what kind of thing they are looking for in search engines. In this case this will help a lot. Going in advance Yahoo! (Glue Pages) has implemented result providing most information in a single page (News, blogs, images, videos, Yahoo answers, sponsored links).

    This kind of new implementation will attract users of search engines and visit website.

  8. Hi cdsseo,

    I do like the idea of a search engine including images with some of their results – it can make those search results pages more interesting and engaging, and it may just make it easier for people to find what they are looking for within those results.

    As you mention, in many searches, the people performing the search are trying to discover new information rather than visiting sites they’ve seen before or trying to find something they already know a lot about, and pictures associated with pages that show up in search results may be helpful to them.

Comments are closed.