Color-based image retrieval using binary signatures

Abstract

We propose two variations of a new image abstraction technique based on signature bit-strings as well as an appropriate similarity metric for color-based image retrieval. Performance evaluation on a heterogeneous database of 20,000 images demonstrated that the proposed technique outperforms well-known approaches while still saving substantial amount of storage space, making it possible to store/search an image database of reasonable size using a few megabytes of main memory (e.g., 4 Mbytes for 100,000 images).