Comparison of level set models in image segmentation

Abstract

Image segmentation is one of the most important tasks in modern imaging applications, which leads to shape reconstruction, volume estimation, object detection and classification. One of the most popular active segmentation models are level set models which are used extensively as an important category of modern image segmentation technique with many different available models to tackle different image applications. Level sets are designed to overcome the topology problems during the evolution of curves in their process of segmentation while the previous algorithms cannot deal with this problem effectively. As a result there is often considerable investigation into the performance of several level set models for a given segmentation problem. It would therefore be helpful to know the characteristics of a range of level set models before applying to a given segmentation problem. In this paper we review a range of level set models and their application to image segmentation work and explain in detail their properties for practical use. IntroductionImage segmentation is the process of partitioning the image into meaningful regions. It is one of the most important tasks in modern imaging for shape reconstruction, volume estimation, object detection and classification. Many different algorithms have been proposed to solve image segmentation problems. In some applications such as medical image segmentation, highly precise models are needed where strong assumptions and anatomical prior are often imposed on the expected segmentation results. The methods based on statistical theory methods can fail in the presence of noise as the artificial parameters should be considered as well, while a lot of physical phenomenon can be described by partial differential equations (PDEs). Those based on PDE, began with the Snake technique introduced by Kass in 1987 [1]. The snake model is based on minimization of an energy term to halt the growth of evolving contours at edges or boundaries. It locks into edges and lines by taking advantage of image forces. Another popular image segmentation method is the level set, introduced in 1988 by Osher-Sethian [2] to overcome the shortcomings of the Snake method such as its topological problem as well as accurate prior knowledge for their initialization.Since level set models are independent of prior knowledge, they are very robust segmentation models when there is no ground truth available. Level set methods are used extensively and have many different applications. As a result there is often considerable investigation into the performance of several level set methods for a given problem. It would therefore be helpful to know the characteristics of a range of level set methods before applying any to a given segmentation problem. Several review papers on level set segmentation are available but each is focused on one area or specific aspect of imaging applications. Some of these topics are: Region-based algorithms As image segmentation is a growing area of research many different robust s...

Radiomics-Led Monitoring of Non-small Cell Lung Cancer Patients During Radiotherapy

“…These include region based image segmentation in [6], medical imaging in [7] and [8], optimal design and inverse problems in [9], piecewise constant applications in [10], deformable models in [11] and [12]. There are also comprehensive reviews covering more general level set applications in [13], [14], [15] and [16]. What is common in all of these articles is that each path is powerful to a particular application but not for every types of images, which is a problem in complex medical images.…”

Colour image segmentation based on a convex K‐means approach

“…Image segmentation aims at dividing an image of N pixels into K regions with similar characteristics together (edges, intensities, colours or textures). Various models and algorithms have been extensively applied for image segmentation, including level-set methods [5][6][7][8][9][10][11], active contours [12][13][14][15], variational models [16][17][18][19] and clustering methods [20][21][22]. Particularly, many successful methods for image segmentation are based on variational models and clustering methods.…”

Color image segmentation based on a convex K-means approach

“…Image segmentation aims at dividing an image of N pixels into K regions with similar characteristics together (edges, intensities, colors or textures). Various models and algorithms have been extensively applied for image segmentation, including level-set methods [7], [8], [9], [10], [11], [12], [13], active contours [14], [15], [16], [17], variational models [18], [19], [20], [21] and clustering methods [22], [23], [24], etc. Particularly, many successful methods for image segmentation are based on variational models and clustering methods.…”