Compact hash codes and data structures for efficient mobile visual search

In this paper we present an efficient method for mobile visual search that exploits compact hash codes and data structures for visual features retrieval. The method has been tested on a large scale standard dataset of one million SIFT features, showing a retrieval performance comparable or superior to state-of-the-art methods, and a very high efficiency in terms of memory consumption and computational requirements. These characteristics make it suitable for application to mobile visual search, where devices have limited computational and memory capabilities.