Robust observation detection for single object tracking: Deterministic and probabilistic patch-based approaches

Mohd Asyraf Zulkifley, David Rawlinson, Bill Moran

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

In video analytics, robust observation detection is very important as the content of the videos varies a lot, especially for tracking implementation. Contrary to the image processing field, the problems of blurring, moderate deformation, low illumination surroundings, illumination change and homogenous texture are normally encountered in video analytics. Patch-Based Observation Detection (PBOD) is developed to improve detection robustness to complex scenes by fusing both feature- and template-based recognition methods. While we believe that feature-based detectors are more distinctive, however, for finding the matching between the frames are best achieved by a collection of points as in template-based detectors. Two methods of PBOD-the deterministic and probabilistic approaches-have been tested to find the best mode of detection. Both algorithms start by building comparison vectors at each detected points of interest. The vectors are matched to build candidate patches based on their respective coordination. For the deterministic method, patch matching is done in 2-level test where threshold-based position and size smoothing are applied to the patch with the highest correlation value. For the second approach, patch matching is done probabilistically by modelling the histograms of the patches by Poisson distributions for both RGB and HSV colour models. Then, maximum likelihood is applied for position smoothing while a Bayesian approach is applied for size smoothing. The result showed that probabilistic PBOD outperforms the deterministic approach with average distance error of 10.03% compared with 21.03%. This algorithm is best implemented as a complement to other simpler detection methods due to heavy processing requirement.

Original languageEnglish
Pages (from-to)15638-15670
Number of pages33
JournalSensors (Switzerland)
Volume12
Issue number11
DOIs
Publication statusPublished - Nov 2012

Fingerprint

Lighting
Observation
Detectors
Poisson distribution
Maximum likelihood
smoothing
Image processing
Textures
Poisson Distribution
Color
Bayes Theorem
Processing
templates
illumination
blurring
detectors
histograms
complement
image processing
textures

Keywords

  • Histogram intersection
  • Maximum correlation
  • Neyman-Pearson method
  • Patch matching
  • Poisson modelling
  • Tracking observation

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Atomic and Molecular Physics, and Optics
  • Analytical Chemistry
  • Biochemistry

Cite this

Robust observation detection for single object tracking : Deterministic and probabilistic patch-based approaches. / Zulkifley, Mohd Asyraf; Rawlinson, David; Moran, Bill.

In: Sensors (Switzerland), Vol. 12, No. 11, 11.2012, p. 15638-15670.

Research output: Contribution to journalArticle

@article{b9e64d0117d54ffda3850212cf60bcd1,
title = "Robust observation detection for single object tracking: Deterministic and probabilistic patch-based approaches",
abstract = "In video analytics, robust observation detection is very important as the content of the videos varies a lot, especially for tracking implementation. Contrary to the image processing field, the problems of blurring, moderate deformation, low illumination surroundings, illumination change and homogenous texture are normally encountered in video analytics. Patch-Based Observation Detection (PBOD) is developed to improve detection robustness to complex scenes by fusing both feature- and template-based recognition methods. While we believe that feature-based detectors are more distinctive, however, for finding the matching between the frames are best achieved by a collection of points as in template-based detectors. Two methods of PBOD-the deterministic and probabilistic approaches-have been tested to find the best mode of detection. Both algorithms start by building comparison vectors at each detected points of interest. The vectors are matched to build candidate patches based on their respective coordination. For the deterministic method, patch matching is done in 2-level test where threshold-based position and size smoothing are applied to the patch with the highest correlation value. For the second approach, patch matching is done probabilistically by modelling the histograms of the patches by Poisson distributions for both RGB and HSV colour models. Then, maximum likelihood is applied for position smoothing while a Bayesian approach is applied for size smoothing. The result showed that probabilistic PBOD outperforms the deterministic approach with average distance error of 10.03{\%} compared with 21.03{\%}. This algorithm is best implemented as a complement to other simpler detection methods due to heavy processing requirement.",
keywords = "Histogram intersection, Maximum correlation, Neyman-Pearson method, Patch matching, Poisson modelling, Tracking observation",
author = "Zulkifley, {Mohd Asyraf} and David Rawlinson and Bill Moran",
year = "2012",
month = "11",
doi = "10.3390/s121115638",
language = "English",
volume = "12",
pages = "15638--15670",
journal = "Sensors (Switzerland)",
issn = "1424-8220",
publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",
number = "11",

}

TY - JOUR

T1 - Robust observation detection for single object tracking

T2 - Deterministic and probabilistic patch-based approaches

AU - Zulkifley, Mohd Asyraf

AU - Rawlinson, David

AU - Moran, Bill

PY - 2012/11

Y1 - 2012/11

N2 - In video analytics, robust observation detection is very important as the content of the videos varies a lot, especially for tracking implementation. Contrary to the image processing field, the problems of blurring, moderate deformation, low illumination surroundings, illumination change and homogenous texture are normally encountered in video analytics. Patch-Based Observation Detection (PBOD) is developed to improve detection robustness to complex scenes by fusing both feature- and template-based recognition methods. While we believe that feature-based detectors are more distinctive, however, for finding the matching between the frames are best achieved by a collection of points as in template-based detectors. Two methods of PBOD-the deterministic and probabilistic approaches-have been tested to find the best mode of detection. Both algorithms start by building comparison vectors at each detected points of interest. The vectors are matched to build candidate patches based on their respective coordination. For the deterministic method, patch matching is done in 2-level test where threshold-based position and size smoothing are applied to the patch with the highest correlation value. For the second approach, patch matching is done probabilistically by modelling the histograms of the patches by Poisson distributions for both RGB and HSV colour models. Then, maximum likelihood is applied for position smoothing while a Bayesian approach is applied for size smoothing. The result showed that probabilistic PBOD outperforms the deterministic approach with average distance error of 10.03% compared with 21.03%. This algorithm is best implemented as a complement to other simpler detection methods due to heavy processing requirement.

AB - In video analytics, robust observation detection is very important as the content of the videos varies a lot, especially for tracking implementation. Contrary to the image processing field, the problems of blurring, moderate deformation, low illumination surroundings, illumination change and homogenous texture are normally encountered in video analytics. Patch-Based Observation Detection (PBOD) is developed to improve detection robustness to complex scenes by fusing both feature- and template-based recognition methods. While we believe that feature-based detectors are more distinctive, however, for finding the matching between the frames are best achieved by a collection of points as in template-based detectors. Two methods of PBOD-the deterministic and probabilistic approaches-have been tested to find the best mode of detection. Both algorithms start by building comparison vectors at each detected points of interest. The vectors are matched to build candidate patches based on their respective coordination. For the deterministic method, patch matching is done in 2-level test where threshold-based position and size smoothing are applied to the patch with the highest correlation value. For the second approach, patch matching is done probabilistically by modelling the histograms of the patches by Poisson distributions for both RGB and HSV colour models. Then, maximum likelihood is applied for position smoothing while a Bayesian approach is applied for size smoothing. The result showed that probabilistic PBOD outperforms the deterministic approach with average distance error of 10.03% compared with 21.03%. This algorithm is best implemented as a complement to other simpler detection methods due to heavy processing requirement.

KW - Histogram intersection

KW - Maximum correlation

KW - Neyman-Pearson method

KW - Patch matching

KW - Poisson modelling

KW - Tracking observation

UR - http://www.scopus.com/inward/record.url?scp=84870566523&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84870566523&partnerID=8YFLogxK

U2 - 10.3390/s121115638

DO - 10.3390/s121115638

M3 - Article

C2 - 23202226

AN - SCOPUS:84870566523

VL - 12

SP - 15638

EP - 15670

JO - Sensors (Switzerland)

JF - Sensors (Switzerland)

SN - 1424-8220

IS - 11

ER -