Review of Level Set in Image Segmentation
- PDF / 2,355,363 Bytes
- 18 Pages / 595.276 x 790.866 pts Page_size
- 55 Downloads / 206 Views
ORIGINAL PAPER
Review of Level Set in Image Segmentation Zhaobin Wang1 · Baozhen Ma1 · Ying Zhu2 Received: 3 November 2019 / Accepted: 4 July 2020 © CIMNE, Barcelona, Spain 2020
Abstract Level set is one of active contour models, which is good at handling complex topologies and capturing boundary. The level set methods are specially used in image with intensity inhomogeneity, such as medical image, SAR image, etc. There are many methods based on level set, which are classified into region-based and edge-based. This article firstly derives the function of curve evolution and original model of level set based on region and edge, respectively. Level set methods over the past decade are summed up and categorized. Some typical models and their improvement are introduced in detail. Some level set methods are employed for comparison. The disadvantages and future work are also discussed.
1 Introduction In the fields of computer vision and image processing, firstly, the objects of interest are separated from the background images that obtained from medical image, SAR, natural image, photograph, etc. Hence, the accuracy of segmentation is the crucial step that influences the validity in next steps of image processing. It is recognized that the level set curve evolution was proposed by Osher and Sethian[1], first employed to solve the problems of crystal growth and flame propagation, and the level set equation was derived from Hamilton–Jacobi equations with parabolic right-hand sides by regulation from hyperbolic conservation laws. In 1993, Caselles et al.[2] introduced curve evolution as a new active contour model into image processing, and gave the relevant mathematical derivation and applications in some simple images. Then almost 30 years later, Malladi et al.[3] used level set to process medical images. Compared with Snake model, its equation is a basic and permanent part of curve motion, and
* Zhaobin Wang [email protected] * Ying Zhu [email protected] 1
School of Infomation Science and Engineering, Lanzhou University, Lanzhou 730000, China
Key Laboratory of Microbial Resources Exploitation and Application of Gansu Province, Institute of Biology, Gansu Academy of Sciences, Lanzhou, China
2
which is stable (satisfied with the maximum principle) and can perform rigorous mathematical analysis. In 2001, Chan and Vese[4] proposed an edge-based level set model on techniques of level sets in 2-D, Mumford–Shah functional and curve evolution, and in 2002[5], they optimized the algorithm based on The Four-Color Theorem and used it in 3-D and more complex images. Then the level set methods are used to segment images with texture, color, movement and shape[6]. Because of the advantage of level set, more and more experts and scholars begin to study image segmentation based on the level set and its optimization. For example, Li et al.[7] proposed a method to process the MR image with intensity inhomogeneity. Wang et al.[8] optimized the Chan–Vese model. Fitting energy thought was introduced by Wang et al.[9,
Data Loading...
