A survey of level set method for image segmentation with intensity inhomogeneity
- PDF / 4,509,260 Bytes
- 25 Pages / 439.642 x 666.49 pts Page_size
- 40 Downloads / 216 Views
A survey of level set method for image segmentation with intensity inhomogeneity Haiping Yu1,2 · Fazhi He1 · Yiteng Pan1 Received: 11 September 2019 / Revised: 18 May 2020 / Accepted: 9 July 2020 / © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Image segmentation is a fundamental task in computer vision and image processing. Due to the presence of high noise, low resolution and intensity inhomogeneity, it is still a difficult problem in the practical applications. Level set methods have been widely used in image processing and computer vision. During the past decades, many models based on level set methods have been proposed to solve image segmentation with intensity inhomogeneity. It is necessary to conduct a comprehensive review and comparison of these models. Specifically, level set methods can be categorized into two groups, including edge-based level set methods (EBLSMs) and region-based level set methods (RBLSMs). This paper reviews some of the recent advances in EBLSMs and RBLSMs for segmenting image with intensity inhomogeneity. Their advantages and disadvantages are discussed in an objective point of view, and their performance is compared on image segmentation with intensity inhomogeneity. Finally, this paper further explores and discusses some open questions in segmenting images with intensity inhomogeneity. Keywords Intensity inhomogeneity · Level set method · Image segmentation · Computer vision
1 Introduction Image segmentation is a fundamental problem in the field of image processing and computer vision [3, 6, 7, 18, 28, 44, 46, 50, 56, 75, 86, 87, 93], and it also plays an important Fazhi He
[email protected] Haiping Yu [email protected] Yiteng Pan [email protected] 1
School of Computer Science, Wuhan University, Wuhan, China
2
School of Computer Science, Huanggang Normal University, Huanggang, China
Multimedia Tools and Applications
role in the medical image processing. The quality of the segmentation will directly affect the results of medical diagnosis. Researchers have proposed many methods, including clustering based methods [1, 16, 80],learning based methods [29, 72, 77, 81, 95] and active contour models(ACMs) . Recently, ACMs are widely used in the field of image segmentation. It not only can provide closed and smooth contours during the curve evolution but also has strong theoretical support. Level set method (LSM), proposed by Osher and Sethian, is one of typical active contour models. The main idea of the level set method is to represent the contour as a zero level set of the higher dimensional function, and convert the motion of the contour into the evolution of the level set function. The zero level set contour at the end of the evolution is the result of the segmentation. Generally, the ACMs are divided into two categories, they are edge-based models (EBMs) [10, 27, 30, 78, 83, 92] and region-base models (RBMs) [11, 22, 48, 59, 67, 73], respectively. EBMs usually use image gradient information to drive the closed contour towards the object bou
Data Loading...
