The Community for Technology Leaders
RSS Icon
Subscribe
pp: 1
Tao Guan , Huazhong University of Science & Technology, wuhan
yunfeng he , Huazhong University of Science & Technology, wuhan
liya duan , Huazhong University of Science & Technology, wuhan
juan gao , Huazhong University of Science & Technology, wuhan
jianzhong yang , Huazhong University of Science & Technology, wuhan
junqing yu , Huazhong University of Science & Technology, wuhan
ABSTRACT
Image descriptor generation and compression are two indispensable steps for on-device Mobile Visual Location Recognition (MVLR) applications. Existing MVLR applications typically rely on Bag-of-Features (BOF) representation which shows superior performance in retrieval accuracy. While promising, the BOF framework will take a large amount of memory and is not compact enough for on-device MVLR applications. In view of that, we make following contributions to the design of a BOF based on-device MVLR system. Firstly, we design an efficient approximate nearest neighbor search algorithm by combining Tree Structured Residual Vector Quantization (TSRVQ) and Residual Vector Quantization (RVQ) for the use of BOF generation purpose. Secondly, we implement a GPS and heading aware RankBoost algorithm to reduce the dimensionality of the generated BOF descriptors. We have evaluated the effectiveness of the proposed algorithms on a HTC mobile phone, with application to on-device mobile visual location recognition in city scale workspace.
CITATION
Tao Guan, yunfeng he, liya duan, juan gao, jianzhong yang, junqing yu, "Efficient Bag-of-Features Generation and Compression for On-Device Mobile Visual Location Recognition", IEEE MultiMedia, , no. 1, pp. 1, PrePrints PrePrints, doi:10.1109/MMUL.2013.31
18 ms
(Ver 2.0)

Marketing Automation Platform Marketing Automation Tool