Lars Iselid of Internetbrus takes a look at search engines for finding pictures and for searching the content of sound and video files.

Magnus Höij, editor-in-chief of the Swedish Internet magazine Internetworld writes in the first 2007 issue that “So far we have no tools for finding relevant information in pictures, sound or films. So far we cannot search for information by entering one of our own pictures in order to find similar information in other images.”

I would say that we do have tools of this kind.

Take for instance a look at Podzinger, Podscope and Blinx. They analyze the spoken words found in sound files and video and make the information searchable.

There are video search engines that index text found in close captioning (subtitles for the hearing impaired).

The picture search service Cobion used to have a function for searching for text displayed in images using OCR technology (optical character recognition).

Findsounds, which has been around since October 2000, can search for sound types. They call their technology content-based audio retrieval.

Both Polar Rose, ImBrowse, Cydral and Riya are working on face recognition techniques.

ImBrowse calls, for instance, its technology for content-based image retrieval. Its image analysis is based on color, structure and form.

Riya has something they call Likness search. If Riya finds that an object in an image is shaped like a handbag, it will search for other images with handbags in them.

If you register an account at Riya, you may upload pictures. Riya will recognize pictures with faces in them and will try to search for with similar faces if you take the time to train the search engine and tag pictures.

So there are actually image and sound search engines out there, in an official version (Cydral, Polar Rose), in a beta version (Riya) or at another stage of development(Cobion). So we are getting there, Mr. Höij!

See also Pandia’s article on the visual shopping search engine Like.com