Revision history [back]

Yes, you need to have a global image descriptor, you can form it by means of Bag of (visual) Words from your local features, see http://answers.opencv.org/question/8677/image-comparison-with-a-database/#8686 or using directly a global image descriptor like GIST, or use even simpler approaches, see the ideas of @StevenPuttemans http://answers.opencv.org/question/9271/global-image-feature-implementation/.

P.S.: Please use the search of this forum next time.