Automatic recognition of biological particles in microscopic images

M. Ranzato, P. E. Taylor, J. M. House, R. C. Flagan, Yann LeCun, P. Perona

Research output: Contribution to journalArticle

Abstract

A simple and general-purpose system to recognize biological particles is presented. It is composed of four stages: First (if necessary) promising locations in the image are detected and small regions containing interesting samples are extracted using a feature finder. Second, differential invariants of the brightness are computed at multiple scales of resolution. Third, after point-wise non-linear mappings to a higher dimensional feature space, this information is averaged over the whole region thus producing a vector of features for each sample that is invariant with respect to rotation and translation. Fourth, each sample is classified using a classifier obtained from a mixture-of-Gaussians generative model. This system was developed to classify 12 categories of particles found in human urine; it achieves a 93.2% correct classification rate in this application. It was subsequently trained and tested on a challenging set of images of airborne pollen grains where it achieved an 83% correct classification rate for the three categories found during one month of observation. Pollen classification is challenging even for human experts and this performance is considered good.

Original languageEnglish (US)
Pages (from-to)31-39
Number of pages9
JournalPattern Recognition Letters
Volume28
Issue number1
DOIs
StatePublished - Jan 1 2007

Fingerprint

Luminance
Classifiers

Keywords

  • Biological particles
  • Cells
  • Feature
  • Mixture of Gaussians (MoG)
  • Non-linearity
  • Pollen
  • Recognition

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Automatic recognition of biological particles in microscopic images. / Ranzato, M.; Taylor, P. E.; House, J. M.; Flagan, R. C.; LeCun, Yann; Perona, P.

In: Pattern Recognition Letters, Vol. 28, No. 1, 01.01.2007, p. 31-39.

Research output: Contribution to journalArticle

Ranzato, M. ; Taylor, P. E. ; House, J. M. ; Flagan, R. C. ; LeCun, Yann ; Perona, P. / Automatic recognition of biological particles in microscopic images. In: Pattern Recognition Letters. 2007 ; Vol. 28, No. 1. pp. 31-39.
@article{2534681e11c6491287a8edde9bf2eeb7,
title = "Automatic recognition of biological particles in microscopic images",
abstract = "A simple and general-purpose system to recognize biological particles is presented. It is composed of four stages: First (if necessary) promising locations in the image are detected and small regions containing interesting samples are extracted using a feature finder. Second, differential invariants of the brightness are computed at multiple scales of resolution. Third, after point-wise non-linear mappings to a higher dimensional feature space, this information is averaged over the whole region thus producing a vector of features for each sample that is invariant with respect to rotation and translation. Fourth, each sample is classified using a classifier obtained from a mixture-of-Gaussians generative model. This system was developed to classify 12 categories of particles found in human urine; it achieves a 93.2{\%} correct classification rate in this application. It was subsequently trained and tested on a challenging set of images of airborne pollen grains where it achieved an 83{\%} correct classification rate for the three categories found during one month of observation. Pollen classification is challenging even for human experts and this performance is considered good.",
keywords = "Biological particles, Cells, Feature, Mixture of Gaussians (MoG), Non-linearity, Pollen, Recognition",
author = "M. Ranzato and Taylor, {P. E.} and House, {J. M.} and Flagan, {R. C.} and Yann LeCun and P. Perona",
year = "2007",
month = "1",
day = "1",
doi = "10.1016/j.patrec.2006.06.010",
language = "English (US)",
volume = "28",
pages = "31--39",
journal = "Pattern Recognition Letters",
issn = "0167-8655",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - Automatic recognition of biological particles in microscopic images

AU - Ranzato, M.

AU - Taylor, P. E.

AU - House, J. M.

AU - Flagan, R. C.

AU - LeCun, Yann

AU - Perona, P.

PY - 2007/1/1

Y1 - 2007/1/1

N2 - A simple and general-purpose system to recognize biological particles is presented. It is composed of four stages: First (if necessary) promising locations in the image are detected and small regions containing interesting samples are extracted using a feature finder. Second, differential invariants of the brightness are computed at multiple scales of resolution. Third, after point-wise non-linear mappings to a higher dimensional feature space, this information is averaged over the whole region thus producing a vector of features for each sample that is invariant with respect to rotation and translation. Fourth, each sample is classified using a classifier obtained from a mixture-of-Gaussians generative model. This system was developed to classify 12 categories of particles found in human urine; it achieves a 93.2% correct classification rate in this application. It was subsequently trained and tested on a challenging set of images of airborne pollen grains where it achieved an 83% correct classification rate for the three categories found during one month of observation. Pollen classification is challenging even for human experts and this performance is considered good.

AB - A simple and general-purpose system to recognize biological particles is presented. It is composed of four stages: First (if necessary) promising locations in the image are detected and small regions containing interesting samples are extracted using a feature finder. Second, differential invariants of the brightness are computed at multiple scales of resolution. Third, after point-wise non-linear mappings to a higher dimensional feature space, this information is averaged over the whole region thus producing a vector of features for each sample that is invariant with respect to rotation and translation. Fourth, each sample is classified using a classifier obtained from a mixture-of-Gaussians generative model. This system was developed to classify 12 categories of particles found in human urine; it achieves a 93.2% correct classification rate in this application. It was subsequently trained and tested on a challenging set of images of airborne pollen grains where it achieved an 83% correct classification rate for the three categories found during one month of observation. Pollen classification is challenging even for human experts and this performance is considered good.

KW - Biological particles

KW - Cells

KW - Feature

KW - Mixture of Gaussians (MoG)

KW - Non-linearity

KW - Pollen

KW - Recognition

UR - http://www.scopus.com/inward/record.url?scp=33751065418&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33751065418&partnerID=8YFLogxK

U2 - 10.1016/j.patrec.2006.06.010

DO - 10.1016/j.patrec.2006.06.010

M3 - Article

AN - SCOPUS:33751065418

VL - 28

SP - 31

EP - 39

JO - Pattern Recognition Letters

JF - Pattern Recognition Letters

SN - 0167-8655

IS - 1

ER -