Research Projects

Abstract: A tenet of object classification is that accuracy improves with an increasing number (and variety) of spectral channels available to the classifier. Hyperspectral images provide hundreds of narrowband measurements over a wide spectral range, and offer superior classification performance over color images. However, hyperspectral data is highly redundant. In this paper we suggest that only 6 measurements are needed to obtain classification results comparable to those realized using hyperspectral data. We present classification results for a natural scene using three imaging modalities: 1) using three broadband color filters (RGB) and three narrowband samples, 2) using six narrowband samples, and 3) using six commonly available optical filters. If these results hold for larger datasets of natural images, recently proposed multispectral image sensors can be used to offer material classification results equal to that of hyperspectral data.

Paper Links:

Click here to download the high resolution (5.3 MB) or low resolution (511 kB) version of the paper.
Or find the paper on IEEE Xplore.

Citation:

J. Holloway, T. Priya, A. Veeraraghavan, and S. Prasad. Image classification in natural scenes: Are a few selective spectral channels suffcient? In Image Processing (ICIP), 2014 IEEE International Conference on, pages 655{659, Oct 2014.

[+] BibTex Citation

@INPROCEEDINGS{holloway2014image,
author={Holloway, J. and Priya, T. and Veeraraghavan, A. and Prasad, S.},
booktitle={Image Processing (ICIP), 2014 IEEE International Conference on},
title={Image Classification in Natural Scenes: Are a Few Selective Spectral Channels Sufficient?},
year={2014},
pages={655-659},
doi={10.1109/ICIP.2014.7025131},
month={Oct},
}

Slides

Download the (smaller) overview presentation accompaning this project as a PowerPoint slideshow (923 kB) or PDF (764 kB).
Download the (larger) presentation accompaning this project as a PowerPoint slideshow (5.3 MB) or PDF (1.1 MB).