Next Generation of Image Management Technologies Adds a New Value to Digital Images

An image is not just a file with a number of technical attributes. It delivers content that until now was considered a single and indivisible unit. The only way to describe image content was defining keywords and descriptions referring to the whole image. With the FotoML standard, the situation can be changed dramatically. It enables storing semantic information about individual objects of image content so providing many new opportunities to search engine companies, photographers, Web site owners, and Web surfers.

There are obvious negative implications of inability to handle individual objects of image content. First, it is impossibility to find a specific object rather than image file. Even if the object is captured by using its name in a list of keywords, the search result is still a file without any clear indication where the desired object is located in the picture. Just imagine that you are looking for David Beckham and get a group photo of Real Madrid without any identification captions. Until you exactly know how he looks, there is no way to find his face on the photo.

Second, an image owner does not have an opportunity to refer to an individual object either. If there is a need to identify people on a group photo or annotate elements of a compound mechanism, the only way is recording element titles in the list of keywords when defining metadata for the image. However, how can one clearly associate a keyword with the corresponding object on the picture?

If such a picture is published on the Web, an ability to refer to individual objects becomes crucial. Since there are millions of other pictures with similar content, chances to find it increases dramatically with individual elements being described providing a possibility to define search request much more specifically.

Finally, if an individual element of an image should be linked to an external resource, such as a Web site or other image, the best solution which currently available approaches can offer is recording the link in a tag which refers to the whole image. However, it will link the whole image with the specified resource rather than a specific object. If there are multiple links, how can one clearly identify to which object exactly each URL corresponds?

Historically, an ability to work with an image on a more granular level referring to specific elements rather than to the whole image challenged people at all times. Various approaches that allowed assigning textual comments to particular objects were known to ancient civilizations. In our days, image annotation is widely used to identify people on photos, provide information which is not explicitly presented on the image, record explanatory notes, and draw viewerвЂ™s attention to important details (read more on different image annotation approaches).

Cogitum LC offers a standardized way to label image content on a lower granular level and work with individual objects presented on an image rather than with the image as a whole.

Cogitum developed an XML-based standard, called FotoML, which allows storing information about individual elements visualized on an image or its parts in an ordinary JPEG file. While the standard is invisible for end-users and does not impose any limitations, such as using specific vocabulary or following a specific procedure, FotoML can be available for search engines.

FotoML offers the following features:

Storing multiple annotations describing individual objects. For each annotation, a text and its position on the image are stored.

Storing information about regions selected on the image and callouts that connect an annotation with an object.

Storing semantically different kinds of annotations in separate layers. For example, on a photo where an architectural ensemble is shown on the background and a group of people is captured on the foreground, annotations that describe the ensemble can be separated from the notes identifying people. So a mechanism for switching between different layers can be provided.

Preserving original position of annotations when cropping or resizing the image.

Supporting captions stored in other formats, such as JPEG UserComment or Exif

Supporting colors of annotation text, background color, and callout color.

Being approved by the search engine companies, FotoML will allow them to display individual objects on images in search results in a similar way as they do it with an ordinary text. Like highlighting a word or phrase that matches the search request, search engines will be able to highlight a specific object on the found image. For example, a user who is looking for Beckham, will get not only a photo of the Real Madrid team, but Beckham's figure highlighted.

To let image owners annotate individual objects on images and store annotations in FotoML, Cogitum offers software tools that enable labeling images located both in the Web and on local computers.

Why FotoML

Search engine companies gain the following benefits:

Image search is not limited by pictures on a photo sharing service which are owned by the search engine company.

Search results include more content meaning an ability to target ads more precisely. Advertisers will appreciate it.

An online annotator can be available as a new service at search engine sites that attract additional traffic.

Image owners enjoy new opportunities:

To make their images available to relevant audience and therefore bring more qualified traffic.

To earn more revenue from placing context ads to their Web pages because ads will be targeted more precisely.

To optimize image content to search engine easily. Annotating an image is very simple and takes from a few seconds to a few minutes.

Web surfers benefit from:

More relevant search results.

More precise search results.

Send Us Your Feedback

We are always happy to get your feedback on the articles. It helps us constantly improve the content and provide you with the most relevant information.

If you know or want to know something that covered or should be mentioned in our articles, please feel free to contact us. We will be glad to get to know your opinion.