An active contour model based on adaptively variable exponent combining Legendre polynomial for image segmentation

Jiajie ZHU, Bin FANG*, Mingliang ZHOU*, Futing LUO, Weizhi XIAN, Gang WANG

*Corresponding author for this work

Research output: Journal PublicationsJournal Article (refereed)peer-review

1 Citation (Scopus)

Abstract

Images with intensity inhomogeneity and blurred boundaries are common in image segmentation tasks, which inevitably result in many difficulties in accurate image segmentation. Massive active contour models (ACMs) have been proposed to solve the problems of intensity inhomogeneity or blurred boundaries respectively. However, there is almost no way to effectively solve the above two problems at the same time, and they are sensitive to the initial contour and noise, or their segmentation speed is relatively slow. In this paper, we propose an active contour model (ACM) based on adaptively variable exponent combining Legendre polynomial (LP) for image segmentation. First, the Legendre polynomial intensity (LPI) is defined, which employs a linear combination of Legendre basis functions for region intensity approximation. Second, an adaptively LPI term is defined, which adopts an adaptively variable exponent function as an acceleration term to drive the curve to quickly evolve to the object boundaries. Third, the distance regularization term is introduced into the active contour as a regularization term to eliminate the need for reinitialization and restrict the behavior of level set function (LSF). Experimental results show that our method offers robustness to gray unevenness, noise and initial curve placement, and adaptability to low contrast and blurred boundaries and outperforms other state-of-the-art algorithms.

Original languageEnglish
Pages (from-to)27495-27522
Number of pages28
JournalMultimedia Tools and Applications
Volume81
Issue number19
Early online date28 Mar 2022
DOIs
Publication statusPublished - Aug 2022
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2022, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.

Keywords

  • Active contour model
  • Image segmentation
  • Legendre polynomial
  • Level set method

Fingerprint

Dive into the research topics of 'An active contour model based on adaptively variable exponent combining Legendre polynomial for image segmentation'. Together they form a unique fingerprint.

Cite this