Retrievr is an experimental image search engine which lets you search and explore in a selection of Flickr photos by drawing a rough sketch (using 4 brushes of different size and color palette) or uploading image to find similar photos. Testing Retrievr I tried a very roughly smile, a firtree and a cup and got a bundle of photos similar to my drawings. The results were fairly fine, sometimes even surprising. Surely it’s only prototype and you can’t demand much of it. But how does it actually work? According to Christian Langreiter, author of this tool, Retrievr creates and stores a compact representation of each photo in its database. The system pulls only the most important features — broad shapes, blocks of color and spatial relationships between different colored areas — out of detailed images to create shorthand approximations of every photo. The storage mechanism extracts the 120 “strongest” features from an image to create something called a “wavelet transform,” which contains much less data than the photo itself and facilitates lightning-fast searches. In this WSJ article you can read more about Retrievr. There is also free offline image search tool ImgSeek for Linux, Windows and Mac OS based on previously mentioned algorithm. The use of ImgSeek is very similar to Retrivr. You simply draw a rough sketch of what you want to find or  just query images similar to one in your collection by double-clicking on it’s thumbnail. You can also group your photos by similarity for easy browsing.