Prague Stringology Conference 2014

Shmuel Tomi Klein and Dana Shapira

Metric Preserving Dense SIFT Compression

Abstract:
The problem of compressing a large collection of feature vectors so that object identification can further be processed on the compressed form of the features is investigated. The idea is to perform matching against a query image in the compressed form of the feature descriptor vectors retaining the metric. Given two SIFT feature vectors, in previous work we suggested to compress them using a lossless encoding for which the pairwise matching can be done directly on the compressed files, by means of a Fibonacci code. In this paper we extend our work to Dense SIFT and in particular to PHOW features, that contain, for each image, about 300 times as many vectors as the original SIFT.

Download article: Article in PostScript Article in PDF BibTeX Reference
 PostScript   PDF   BibTeX reference 
Download presentation: Presentation