Efficient and Distinct Large Scale Bags of Words

Due to the increasing flood of digital images and the overall increase of storage capacity, large scale image databases are common these days. Managing such a vast number of digital images is not trivial. This work deals with the problem of finding replicas in image databases containing more than 100 000 images. The bag of visual words method, in analogy to the bag of words method in text search applications, is examined in detail. Every image can be seen as a set of visual words. Computation of these visual words requires clustering of a huge amount of data in a fast yet accurate way. This is the most time consuming part in such an application. A clustering algorithm is developed that has linear runtime and can be carried out in parallel. We observe that with increasing size of the database, the problem of decreasing discrimination between high frequency images arises. Features of images with natural repetitive texture become similar to other images and show up in most of the search results. This problem is addressed by developing an asymmetric Hamming distance measurement for bags of visual words. It allows better discrimination power in large databases, while being robust to image transformations such as rotation, crop, or change of resolution and size.