Reproducible kernel Hilbert space based global and local image segmentation

Liam Burrows; Weihong Guo; Ke Chen; Francesco Torella

doi:10.3934/ipi.2020048

Article Contents

2021, Volume 15, Issue 1: 1-25. Doi: 10.3934/ipi.2020048

This issue Previous Article Next Article Automatic extraction of cell nuclei using dilated convolutional network

Reproducible kernel Hilbert space based global and local image segmentation

1.
Centre for Mathematical Imaging Techniques and Department of Mathematical Sciences, University of Liverpool, United Kingdom
2.
Department of Mathematics, Applied Mathematics and Statistics, Case Western Reserve, University, Cleveland, OH 44106, USA
3.
Liverpool Vascular & Endovascular Service, Royal Liverpool and Broadgreen University, Hospitals NHS Trust, Liverpool, L7 8XP, United Kingdom

Received: December 2019

Revised: May 2020

Early access: August 2020

Published: February 2021

Abstract / Introduction Full Text(HTML) Figure(10) Related Papers Cited by

Abstract

Image segmentation is the task of partitioning an image into individual objects, and has many important applications in a wide range of fields. The majority of segmentation methods rely on image intensity gradient to define edges between objects. However, intensity gradient fails to identify edges when the contrast between two objects is low. In this paper we aim to introduce methods to make such weak edges more prominent in order to improve segmentation results of objects of low contrast. This is done for two kinds of segmentation models: global and local. We use a combination of a reproducing kernel Hilbert space and approximated Heaviside functions to decompose an image and then show how this decomposition can be applied to a segmentation model. We show some results and robustness to noise, as well as demonstrating that we can combine the reconstruction and segmentation model together, allowing us to obtain both the decomposition and segmentation simultaneously.

Keywords:

Mathematics Subject Classification: Primary: 58F15, 58F17; Secondary: 53C35.

Citation:

Full Text(HTML)

Figure 1. Comparison of Geodesic and Euclidean distance constraints

Download: Full-size image PowerPoint slide

Figure 2. Left panel: 3D surface images of $ \psi $ of nine parameter pairs $ (\theta, c) $; right panel: the corresponding 2D images. In each panel, from left to right and then from top to bottom: $ ( \frac{4 \pi}{5} , \frac{51}{1024} ), ( \frac{4\pi}{5} , \frac{25}{64} ), ( \frac{4\pi}{5} , \frac{175}{256} ), ( \frac{6\pi}{5}, \frac{135}{1024} ), ( \frac{6\pi}{5}, \frac{1}{2} ), ( \frac{6\pi}{5}, \frac{25}{32} ), ( \frac{8\pi}{5}, \frac{5}{64} ) $, $ ( \frac{8\pi}{5}, \frac{75}{256} ), ( \frac{8\pi}{5}, \frac{75}{128} ) $

Download: Full-size image PowerPoint slide

Figure 3. The decomposition of the input image (top left), into the smooth parts (top middle) and edge parts (top right), obtained from RKHS model (8), and edge indicator function $ g(\Psi \beta) $ (bottom left). In comparison, we display the gradient of the image, $ |\nabla z| $ and the edge indicator function $ g(|\nabla z|) $ in the bottom middle and bottom right respectively

Download: Full-size image PowerPoint slide

Figure 4. Segmentation of white matter from MRI brain image. We show both the contour overlaid on the original image, and the associated binary segmentation result for each example. First column: segmentation using edge weighted Chan-Vese using $ |\nabla z| $ model (24). Second column: Edge weighted Chan-Vese model (24) using Canny ($ g = g_{\mathcal{C}} $). Third column: Cai et al. [9]. Fourth column: Proposed two stage global model (8) and (25). Final column: The proposed combined global segmentation model (26)

Download: Full-size image PowerPoint slide

Figure 5. Row 1: 3% noise, Row 2: 5% noise, Row 3: 7% noise, Row 4: 9% noise. We show both the contour overlaid on the original image, and the associated binary segmentation result for each example. First two columns: Using model (24). Final two columns: Combined model, algorithm (1)

Download: Full-size image PowerPoint slide

Figure 6. Geodesic distances computed using gradient (middle) and $ \Psi \beta $ (right)

Download: Full-size image PowerPoint slide

Figure 7. Row one: Original clean image and the user input. Row two: Image and result with 20% noise. Row three: Image and result with 40% noise. Row four: Image and result with 60% noise

Download: Full-size image PowerPoint slide

Figure 8. Segmentation of blood vessel

Download: Full-size image PowerPoint slide

Figure 9. Segmentation of blood vessel

Download: Full-size image PowerPoint slide

Figure 10. Result of the combined selective model

Download: Full-size image PowerPoint slide

Related Papers

Cited by

References

[1]	L. Ambrosio and V. Tortorelli, On the approximation of free discontinuity problems, Bollettino Dell'Unione Matematica Italiana. B, 6 (1992), 105-123.
[2]	L. Ambrosio and V. M. Tortorelli, Approximation of functional depending on jumps by elliptic functional via t-convergence, Communications on Pure and Applied Mathematics, 43 (1990), 999-1036. doi: 10.1002/cpa.3160430805.
[3]	B. Appleton and H. Talbot, Globally minimal surfaces by continuous maximal flows, IEEE Transactions on Pattern Analysis and Machine Intelligence, 28 (2006), 106-118. doi: 10.1109/TPAMI.2006.12.
[4]	N. Badshah and K. Chen, Image selective segmentation under geometrical constraints using an active contour approach, Communications in Computational Physics, 7 (2010), 759-778. doi: 10.4208/cicp.2009.09.026.
[5]	E. Bae, J. Yuan and X.-C. Tai, Global minimization for continuous multiphase partitioning problems using a dual approach, International Journal of Computer Vision, 92 (2011), 112-129. doi: 10.1007/s11263-010-0406-y.
[6]	Y. Boykov and V. Kolmogorov, An experimental comparison of min-cut/max-flow algorithms for energy minimization in vision, Energy Minimization Methods in Computer Vision and Pattern Recognition, 2134 (2001), 359-374. doi: 10.1007/3-540-44745-8_24.
[7]	Y. Boykov, O. Veksler and R. Zabih, Fast approximate energy minimization via graph cuts, Proceedings of the Seventh IEEE International Conference on Computer Vision, 1 (1999), 377-384. doi: 10.1109/ICCV.1999.791245.
[8]	X. Bresson, S. Esedoḡlu, P. Vandergheynst, J.-P. Thiran and S. Osher, Fast global minimization of the active contour/snake model, Journal of Mathematical Imaging and Vision, 28 (2007), 151-167. doi: 10.1007/s10851-007-0002-0.
[9]	X. Cai, R. Chan and T. Zeng, A two-stage image segmentation method using a convex variant of the Mumford–Shah model and thresholding, SIAM Journal on Imaging Sciences, 6 (2013), 368-390. doi: 10.1137/120867068.
[10]	V. Caselles, R. Kimmel and G. Sapiro, Geodesic active contours, International Journal of Computer Vision, 22 (1997), 61-79. doi: 10.1109/ICCV.1995.466871.
[11]	A. Chambolle, An algorithm for total variation minimization and applications, Journal of Mathematical Imaging and Vision, 20 (2004), 89-97.
[12]	T. F. Chan, S. Esedoglu and M. Nikolova, Algorithms for finding global minimizers of image segmentation and denoising models, SIAM Journal on Applied Mathematics, 66 (2006), 1632-1648. doi: 10.1137/040615286.
[13]	T. F. Chan and L. A. Vese, Active contours without edges, IEEE Transactions on Image Processing, 10 (2001), 266-277. doi: 10.1109/83.902291.
[14]	L.-J. Deng, W. Guo and T.-Z. Huang, Single-image super-resolution via an iterative reproducing kernel hilbert space method, IEEE Transactions on Circuits and Systems for Video Technology, 26 (2016), 2001-2014. doi: 10.1109/TCSVT.2015.2475895.
[15]	C. Gout, C. Le Guyader and L. Vese, Segmentation under geometrical conditions using geodesic active contours and interpolation using level set methods, Numerical Algorithms, 39 (2005), 155-173. doi: 10.1007/s11075-004-3627-8.
[16]	J.-B. Hiriart-Urruty and C. Lemar{é}chal, Convex Analysis and Minimization Algorithms I: Fundamentals, volume 305., Springer Science & Business Media, 2013.
[17]	P. Jaccard, The distribution of flora in the apline zone.1, New Phytologist, 11 (1912), 37-50.
[18]	S. H. Kang, B. Shafei and G. Steidl, Supervised and transductive multi-class segmentation using p-Laplacians and RKHS methods, Journal of Visual Communication and Image Representation, 25 (2014), 1136-1148. doi: 10.1016/j.jvcir.2014.03.010.
[19]	M. Kass, A. Witkin and D. Terzopoulos, Snakes: Active contour models, International Journal of Computer Vision, 1 (1988), 321-331. doi: 10.1007/BF00133570.
[20]	C. Li, C.-Y. Kao, J. C. Gore and Z. Ding, Minimization of region-scalable fitting energy for image segmentation, IEEE Transactions on Image Processing, 17 (2008), 1940-1949. doi: 10.1109/TIP.2008.2002304.
[21]	C. Liu, M. Ng and T. Zeng, Weighted variational model for selective image segmentation with application to medical images, Pattern Recognition, 76 (2018), 367-379. doi: 10.1016/j.patcog.2017.11.019.
[22]	Z. Li and F. Malgouyres, Regularized non-local total variation and application in image restoration, Journal of Mathematical Imaging and Vision, 59 (2017), 296-317. doi: 10.1007/s10851-017-0732-6.
[23]	D. Mumford and J. Shah, Optimal approximations by piecewise smooth functions and associated variational problems, Communications on Pure and Applied Mathematics, 42 (1989), 577-685. doi: 10.1002/cpa.3160420503.
[24]	S. Osher and J. A. Sethian, Fronts propagating with curvature-dependent speed: Algorithms based on hamilton-jacobi formulations, Journal of Computational Physics, 79 (1988), 12-49. doi: 10.1016/0021-9991(88)90002-2.
[25]	R. B. Potts, Some generalized order-disorder transformations, Mathematical Proceedings of the Cambridge Philosophical Society, 48 (1952), 106-109. doi: 10.1017/S0305004100027419.
[26]	M. Roberts, K. Chen and K. L. Irion, A convex geodesic selective model for image segmentation, Journal of Mathematical Imaging and Vision, 61 (2019), 482-503. doi: 10.1007/s10851-018-0857-2.
[27]	M. Roberts and J. Spencer, Chan–vese reformulation for selective image segmentation, Journal of Mathematical Imaging and Vision, 61 (2019), 1173-1196. doi: 10.1007/s10851-019-00893-0.
[28]	J. Spencer and K. Chen, A convex and selective variational model for image segmentation, Communications in Mathematical Sciences, 13 (2015), 1453-1472. doi: 10.4310/CMS.2015.v13.n6.a5.
[29]	X.-F. Wang, D.-S. Huang and H. Xu, An efficient local chan–vese model for image segmentation, Pattern Recognition, 43 (2010), 603-618. doi: 10.1016/j.patcog.2009.08.002.
[30]	Y. Xu and W. Yin, A block coordinate descent method for regularized multiconvex optimization with applications to nonnegative tensor factorization and completion, SIAM Journal on Imaging Sciences, 6 (2013), 1758-1789. doi: 10.1137/120887795.
[31]	J. Yuan, E. Bae, X.-C. Tai and Y. Boykov, A continuous max-flow approach to potts model, In European Conference on Computer Vision, Springer, 2010,379–392.
[32]	J. Yuan, E. Bae, X.-C. Tai and Y. Boykov, A spatially continuous max-flow and min-cut framework for binary labeling problems, Numerische Mathematik, 126 (2014), 559-587. doi: 10.1007/s00211-013-0569-x.
[33]	H. Zhang, Y. Chen and J. Shi, Nonparametric image segmentation using Renyi's statistical dependence measure, Journal of Mathematical Imaging and Vision, 44 (2012), 330-340. doi: 10.1007/s10851-012-0329-z.
[34]	H. Zhao, A fast sweeping method for Eikonal equations, Mathematics of Computation, 74 (2005), 603-627. doi: 10.1090/S0025-5718-04-01678-3.