Image similarity search with compact data structures

Abstract

The recent theoretical advances on compact data structures (also called "sketches") have raised the question of whether they can effectively be applied to content-based image retrieval systems. The main challenge is to derive an algorithm that achieves high-quality similarity searches while using compact metadata. This paper proposes a new similarity search method consisting of three parts. The first is a new region feature representation with weighted L1 distance function, and EMD* match, an improved EMD match, to compute image similarity. The second is a thresholding and transformation algorithm to convert feature vectors into very compact data structures. The third is an EMD embedding based filtering method to speed up the query process. We have implemented a prototype system with the proposed method and performed experiments with a 10,000 image database. Our results show that the proposed method can achieve more effective similarity searches than previous approaches with metadata 3 to 72 times more compact than previous systems. The experiments also show that our EMD embedding based filtering technique can speed up the query process by a factor of 5 or more with little loss in query effectiveness.

Image Similarity Search using a Negative Selection Algorithm

“…There are many more possibilities notably measures based on Earth Mover's Distance, which is a similarity measure between multidimensional distributions (Lv et al, 2004). Other similarity measures include contrast measurements, color matrices or even completely different approaches such as edge detection.…”

Design patterns for tunable and efficient SSD-based indexes

“…LSH-based index. Locality sensitive hashing [24] is a technique used in the multimedia community [26,9] for finding duplicate videos and images at large scale. LSH systems use multiple hash tables.…”

String Distances for Near-duplicate Detection

“…On the problem of near-duplicate image detection, [6] applied compact data structure to region-based image retrieval using EMD (Earth Mover's Distance) and compared their results positively with previous systems. [7] have applied the neuroscience-inspired Visual Attention Similarity Measure in order to give more weight to regions of interest.…”