Rus Articles Journal

How to find the picture in the Internet?

How many times happened that, having found some photo on the disk, you could not remember - from where it undertook? Or could not understand at all what on it is represented? And how many pictures it is necessary to shovel searching “something here it, but it is a little of another“? Without speaking about immemorial requirement to find the original of the photo on its miniature.

Now new search service from Google will help to solve this problem. In addition to search of pictures now Google allows to carry out search on to pictures and photos. At the same time the available image is used as a sample for search. If you work in the Google Chrome browser, then can just drag the picture in a search line, and Google will issue information on all websites on which there is such picture. Besides, Google will offer the most probable name for this image. Owners of the Chrome and Firefox browsers can also get plug-ins which allow to carry out search by pressing of the right button of a mouse on the initial picture.

It is necessary to tell that this idea is not revolutionary. Since 2008 there is TinEye - the search engine specializing in search of images on the Internet. TinEye became famous for the fact that with its help according to the photo the personality of the unknown soldier was identified. The system managed to find the file with the complete list of the people represented on a photo. But, despite age of TinEye, quite respectable for the website, young service from Google has stronger algorithms of search and more volume base of images already now. Besides, unlike TinEye, Google offers not only completely coinciding, but also similar pictures that, undoubtedly, will be to the taste to web - designers.

Technically process of search of images almost does not differ from usual text search. Robots of Google bypass the World wide web and index the pictures met on the way and photos. Of course, creation of an index of images is much more difficult, than indexation of text documents - for each picture characteristic elements which then are described definitely are allocated, are sorted and remain in the database. Thus, each image receives the unique key.

In the course of search the sample loaded by the user is also presented in the form of a key by which there is a comparison with the keys which are already available in the database. It is natural that full coincidence is possible, only if photos are absolutely identical. However the used way of the description allows to recognize similar images irrespective of the format of the file, the size, chromaticity and other characteristics which are not influencing sense of the image. The identification even as regards the initial picture is quite possible.

Of course, all this is possible only if on the Internet everything is there is a similar photo and there is its description. So for personal photos it is still possible not to be afraid. But if you like to load pages on social networks with the photos, then can not worry - soon Google and about you learns.

For example, the system identified Andrey Arshavin and Madonna`s photo absolutely surely. The only thing that could confuse Google is the specularly displayed image. But the first step is the hardest, algorithms of search will be improved.