Feature extraction techniques for exploratory visualization of vector-valued imagery

G. Harikumar, Yoram Bresler

Research output: Contribution to journalReview article

Abstract

This paper addresses the exploratory visualization of multispectral image data. In such data, each component of the vector pixel corresponds to a different imaging modality or a different combination of imaging parameters, and may provide different levels of contrast sensitivity between different regions of the underlying image. We address the problem of presenting this multidimensional data to human observers by synthesizing a display matched to their visual capabilities. Specifically, we seek to determine a data-adaptive linear projection of the vector data to one dimension that produces a grayscale image providing maximum discrimination between the different regions of the underlying object. The approach is equivalent to the extraction of the best linear feature of the vector field. Several new feature-extraction criteria that take into account both the spatial and multivariate structures of the data are proposed and illustrated by simulations on test images.

Original languageEnglish (US)
Pages (from-to)1324-1334
Number of pages11
JournalIEEE Transactions on Image Processing
Volume5
Issue number9
DOIs
StatePublished - Dec 1 1996

Fingerprint

Feature extraction
Visualization
Imaging techniques
Pixels
Display devices

ASJC Scopus subject areas

  • Software
  • Computer Graphics and Computer-Aided Design

Cite this

Feature extraction techniques for exploratory visualization of vector-valued imagery. / Harikumar, G.; Bresler, Yoram.

In: IEEE Transactions on Image Processing, Vol. 5, No. 9, 01.12.1996, p. 1324-1334.

Research output: Contribution to journalReview article

@article{d1b86f88e16744ddb15a51334e74c3d5,
title = "Feature extraction techniques for exploratory visualization of vector-valued imagery",
abstract = "This paper addresses the exploratory visualization of multispectral image data. In such data, each component of the vector pixel corresponds to a different imaging modality or a different combination of imaging parameters, and may provide different levels of contrast sensitivity between different regions of the underlying image. We address the problem of presenting this multidimensional data to human observers by synthesizing a display matched to their visual capabilities. Specifically, we seek to determine a data-adaptive linear projection of the vector data to one dimension that produces a grayscale image providing maximum discrimination between the different regions of the underlying object. The approach is equivalent to the extraction of the best linear feature of the vector field. Several new feature-extraction criteria that take into account both the spatial and multivariate structures of the data are proposed and illustrated by simulations on test images.",
author = "G. Harikumar and Yoram Bresler",
year = "1996",
month = "12",
day = "1",
doi = "10.1109/83.535844",
language = "English (US)",
volume = "5",
pages = "1324--1334",
journal = "IEEE Transactions on Image Processing",
issn = "1057-7149",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "9",

}

TY - JOUR

T1 - Feature extraction techniques for exploratory visualization of vector-valued imagery

AU - Harikumar, G.

AU - Bresler, Yoram

PY - 1996/12/1

Y1 - 1996/12/1

N2 - This paper addresses the exploratory visualization of multispectral image data. In such data, each component of the vector pixel corresponds to a different imaging modality or a different combination of imaging parameters, and may provide different levels of contrast sensitivity between different regions of the underlying image. We address the problem of presenting this multidimensional data to human observers by synthesizing a display matched to their visual capabilities. Specifically, we seek to determine a data-adaptive linear projection of the vector data to one dimension that produces a grayscale image providing maximum discrimination between the different regions of the underlying object. The approach is equivalent to the extraction of the best linear feature of the vector field. Several new feature-extraction criteria that take into account both the spatial and multivariate structures of the data are proposed and illustrated by simulations on test images.

AB - This paper addresses the exploratory visualization of multispectral image data. In such data, each component of the vector pixel corresponds to a different imaging modality or a different combination of imaging parameters, and may provide different levels of contrast sensitivity between different regions of the underlying image. We address the problem of presenting this multidimensional data to human observers by synthesizing a display matched to their visual capabilities. Specifically, we seek to determine a data-adaptive linear projection of the vector data to one dimension that produces a grayscale image providing maximum discrimination between the different regions of the underlying object. The approach is equivalent to the extraction of the best linear feature of the vector field. Several new feature-extraction criteria that take into account both the spatial and multivariate structures of the data are proposed and illustrated by simulations on test images.

UR - http://www.scopus.com/inward/record.url?scp=0030242088&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030242088&partnerID=8YFLogxK

U2 - 10.1109/83.535844

DO - 10.1109/83.535844

M3 - Review article

C2 - 18285221

AN - SCOPUS:0030242088

VL - 5

SP - 1324

EP - 1334

JO - IEEE Transactions on Image Processing

JF - IEEE Transactions on Image Processing

SN - 1057-7149

IS - 9

ER -