Dictionary Design Algorithms for Vector Map Compression

Abstract

Vector maps (e.g. road maps) are important in a variety ofapplications including mobile computing. Due to the large size of vector maps, only a small part of maps (e.g. relevant to current location of the vehicle) can be cached in hand-held or in-vehicle devices used for mobile computing. Compression techniques for vector maps can help cache larger subsets of maps and reduce the communicationcosts of downloading newer subsets of maps during travel.Dictionary-based compression technique one common means of data compression. This paper explores the problem of designing dictionaries for dictionary based compression techniques for vector maps. We propose a novel clustering-based dictionary design. The proposed approach adapts the dictionary to a given dataset, yielding better approximation. Experimental evaluation shows that when the dictionary size is fixed, the proposed clustering-based technique achieves better accuracy compared with conventional approaches.