Hardware/software co-design of embedded real-time KD-tree based feature matching systems

Saad Shoaib, Rehan Hafiz, Muhammad Shafique

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Feature matching is an important step in many computational photography applications such as image stitching, 3D reconstruction and object recognition. KD-trees based Best Bin First (KD-BBF) search is one of the most widely used feature matching scheme being employed along with SIFT and SURF. The real time requirements of such computer vision applications for embedded systems put tight compute bounds on the processor. In this paper we propose a soft-core and a hardware accelerator based architecture that enables real time matching of SIFT feature descriptors for HD resolution images at 30 FPS. The proposed accelerator provides a speedup of more than 8 times over the pure software implementation.

Original languageEnglish (US)
Title of host publicationAdvances in Visual Computing - 10th International Symposium, ISVC 2014, Proceedings
EditorsGeorge Bebis, Richard Boyle, Bahram Parvin, Darko Koracin, Ryan McMahan, Jason Jerald, Hui Zhang, Steven M. Drucker, Kambhamettu Chandra, El Choubassi Maha, Zhigang Deng, Mark Carlson
PublisherSpringer Verlag
Pages936-945
Number of pages10
ISBN (Electronic)9783319143637
DOIs
StatePublished - 2014
Event10th International Symposium on Visual Computing, ISVC 2014 - Las Vegas, United States
Duration: Dec 8 2014Dec 10 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8888
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference10th International Symposium on Visual Computing, ISVC 2014
Country/TerritoryUnited States
CityLas Vegas
Period12/8/1412/10/14

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Hardware/software co-design of embedded real-time KD-tree based feature matching systems'. Together they form a unique fingerprint.

Cite this