“Local Rank Differences” Image Feature Implemented on GPU
Brno University of Technology, Faculty of Information Technology, Bozetechova 2, 612 66 Brno, Czech Republic
In Advanced Concepts for Intelligent Vision Systems (2008), pp. 170-181
@conference{polok2008local,
title={“Local Rank Differences” Image Feature Implemented on GPU},
author={Polok, L. and Herout, A. and Zem{v{c}}{‘i}k, P. and Hradi{v{s}}, M. and Jur{‘a}nek, R. and Jo{v{s}}th, R.},
booktitle={Advanced Concepts for Intelligent Vision Systems},
pages={170–181},
year={2008},
organization={Springer}
}
A currently popular trend in object detection and pattern recognition is usage of statistical classifiers, namely AdaBoost and its modifications. The speed performance of these classifiers largely depends on the low level image features they are using: both on the amount of information the feature provides and the executional time of its evaluation. Local Rank Differences is an image feature that is alternative to commonly used haar wavelets. It is suitable for implementation in programmable (FPGA) or specialized (ASIC) hardware, but – as this paper shows – it performs very well on graphics hardware (GPU) as well. The paper discusses the LRD features and their properties, describes an experimental implementation of LRD in graphics hardware, presents its empirical performance measures compared to alternative approaches and suggests several notes on practical usage of LRD and proposes directions for future work.
December 9, 2010 by hgpu
