The relationship between curvilinear structure enhancement and ridge detection methods
- PDF / 5,759,005 Bytes
- 21 Pages / 595.276 x 790.866 pts Page_size
- 50 Downloads / 185 Views
ORIGINAL ARTICLE
The relationship between curvilinear structure enhancement and ridge detection methods Haifa F. Alhasson1,3
· Chris G. Willcocks1 · Shuaa S. Alharbi1,3 · Adetayo Kasim2 · Boguslaw Obara1
Accepted: 19 September 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Curvilinear structure detection and quantification is a large research area with many imaging applications in fields such as biology, medicine, and engineering. Curvilinear enhancement is often used as a pre-processing stage for ridge detection, but there has been little investigation into the relationship between enhancement and ridge detection. In this paper, we thoroughly evaluate the pair-wise combinations of different curvilinear enhancement and ridge detection methods across two highly varied datasets, as well as samples of three other datasets. In particular, we present the approaches complementing one another and the gained insights, which will aid researchers in designing generic ridge detectors. Keywords Curvilinear structures · Ridge detection · Curvilinear enhancement · Skeletonisation · Object detection · Image analysis
1 Introduction Curvilinear structure enhancement and skeletonisation approaches are fundamental tools used in image processing. The enhancement process refers to a set of techniques that seek to improve the interpretability or perception of objects in the image for human observation or to provide better input for another image analysis task, such as ridge detection or segmentation. However, curvilinear skeletonisation helps to describe the shape of curvilinear structures as a single one-pixel-wide path on pre-segmented (binary) images. Ridge detection is considered to be a special case of skeletonisation, where the extraction of the ridge is a free-segmentation process that is calculated directly from the greyscale image. Therefore, both skeletonisation and ridge detection processes can be referred to as line extraction or ridge detection. Many enhancement and line extraction approaches have been discussed over the last several decades,
B
Boguslaw Obara [email protected]
1
Department of Computer Science, Durham University, Durham, UK
2
Department of Anthropology, Durham University, Durham, UK
3
Department of Information Technology, Collage of Computer, Qassim University, Buraidah, Saudi Arabia
and extensive reports have been published [1–5]. Most line extractors were designed for a specific application, combining a line detection step with another line discrimination step using prior or contextual knowledge, such as the length or width of line segment (e.g. [6]). However, the images targeted for investigation in most applications require further refinement, such as noise reduction, in order to successfully extract lines. Therefore, researchers usually combine a prior enhancement step with the line extraction approach. The combination of these approaches is used in a vast number of applications in many domains, including but not limited to the following: – Bioscience
Data Loading...
