# American Institute of Mathematical Sciences

June  2019, 13(3): 653-677. doi: 10.3934/ipi.2019030

## A dual EM algorithm for TV regularized Gaussian mixture model in image segmentation

 1 Laboratory of Mathematics and Complex Systems (Ministry of Education of China), School of Mathematical Sciences, Beijing Normal University, Beijing 100875, China 2 Department of Mathematics, Hong Kong Baptist University, Kowloon Tong, Hong Kong, China 3 Depart of Mathematics, University of Bergen, P. O. Box 7800, N-5020, Bergen, Norway

* Corresponding author: Jun Liu

Received  May 2018 Revised  October 2018 Published  March 2019

A dual expectation-maximization (EM) algorithm for total variation (TV) regularized Gaussian mixture model (GMM) is proposed in this paper. The algorithm is built upon the EM algorithm with TV regularization (EM-TV) model which combines the statistical and variational methods together for image segmentation. Inspired by the projection algorithm proposed by Chambolle, we give a dual algorithm for the EM-TV model. The related dual problem is smooth and can be easily solved by a projection gradient method, which is stable and fast. Given the parameters of GMM, the proposed algorithm can be seen as a forward-backward splitting method which converges. This method can be easily extended to many other applications. Numerical results show that our algorithm can provide high quality segmentation results with fast computation speed. Compared with the well-known statistics based methods such as hidden Markov random field with EM method (HMRF-EM), the proposed algorithm has a better performance. The proposed method could also be applied to MRI segmentation such as SPM8 software and improve the segmentation results.

Citation: Shi Yan, Jun Liu, Haiyang Huang, Xue-Cheng Tai. A dual EM algorithm for TV regularized Gaussian mixture model in image segmentation. Inverse Problems & Imaging, 2019, 13 (3) : 653-677. doi: 10.3934/ipi.2019030
##### References:

show all references

##### References:
Comparison of Dual EM-TV, Original EM-TV [26] and HMRF-EM methods for 2 classes segmentation
Enlarged red square regions in Figure 1
Comparison of Dual EM-TV, Original EM-TV [26] and HMRF-EM methods for 4 classes segmentation
Enlarged red square regions in Figure 3
An image with two regions, which are the same mean but different variance, and the histogram
Comparison of segmentation results of GMM-EM, Dual EM-TV and CV model on region with same mean but different variances image
Comparison between Dual EM-TV and HMRF-EM method
Segmentation results by Dual EM-TV in color images for two-region images
Segmentation results by Dual EM-TV in color images for multi-region images
Segmentation comparison under different noise levels. Noise standard deviation: A: 120/255, B: 240/255, C: 300/255, D: 390/255
Segmentation comparison of the same multi-region image by different noise levels. Noise standard deviation: A: 50/255, B: 60/255, C: 70/255, D: 80/255
Results of SPM8 with one slice of MR Images. First row: left image, MRI with 0% RF and 0% noise; right: MRI with 40% RF and 9% noise). Second and third rows: the first and third columns are the newsegment method for left and right images, respectively; the second and fourth columns are the proposed method for left and right images, respectively
Comparison of 2 classes segmentation for Dual EM-TV, Original EM-TV [26] and HMRF-EM method (Heart image with size $615\times 615$)
 Methods CPU time (seconds) Accuracy (SAI) Dual EM-TV 1.233 99.70% Original EM-TV [26] 9.482 98.02% HMRF-EM 4n 4.052 91.84% HMRF-EM 8n 3.052 97.38% %HMRF-EM 12n 2.653 98.69% HMRF-EM 20n 1.528 99.24%
 Methods CPU time (seconds) Accuracy (SAI) Dual EM-TV 1.233 99.70% Original EM-TV [26] 9.482 98.02% HMRF-EM 4n 4.052 91.84% HMRF-EM 8n 3.052 97.38% %HMRF-EM 12n 2.653 98.69% HMRF-EM 20n 1.528 99.24%
Comparison of 4 classes segmentation for Dual EM-TV, Original EM-TV [26] and HMRF-EM methods (4-color image with size $600\times 600$)
 Methods CPU time (seconds) Accuracy (SAI) Dual EM-TV 1.699 99.90% Original EM-TV [26] 18.502 99.02% HMRF-EM 4n 3.480 97.39% HMRF-EM 8n 3.202 99.81% %HMRF-EM 12n 2.696 99.87% HMRF-EM 20n 2.388 99.83%
 Methods CPU time (seconds) Accuracy (SAI) Dual EM-TV 1.699 99.90% Original EM-TV [26] 18.502 99.02% HMRF-EM 4n 3.480 97.39% HMRF-EM 8n 3.202 99.81% %HMRF-EM 12n 2.696 99.87% HMRF-EM 20n 2.388 99.83%
Comparison of DM in SPM8 on brain images
 noise level 5% 5% 7% 7% 9% 9% brain part WM GM WM GM WM GM Dual EM-TV 0.9380 0.9122 0.9218 0.8970 0.9035 0.8818 New Segment 0.9317 0.9099 0.9035 0.8822 0.8759 0.8536
 noise level 5% 5% 7% 7% 9% 9% brain part WM GM WM GM WM GM Dual EM-TV 0.9380 0.9122 0.9218 0.8970 0.9035 0.8818 New Segment 0.9317 0.9099 0.9035 0.8822 0.8759 0.8536
 [1] Bao Wang, Alex Lin, Penghang Yin, Wei Zhu, Andrea L. Bertozzi, Stanley J. Osher. Adversarial defense via the data-dependent activation, total variation minimization, and adversarial training. Inverse Problems & Imaging, 2021, 15 (1) : 129-145. doi: 10.3934/ipi.2020046 [2] Liam Burrows, Weihong Guo, Ke Chen, Francesco Torella. Reproducible kernel Hilbert space based global and local image segmentation. Inverse Problems & Imaging, 2021, 15 (1) : 1-25. doi: 10.3934/ipi.2020048 [3] Balázs Kósa, Karol Mikula, Markjoe Olunna Uba, Antonia Weberling, Neophytos Christodoulou, Magdalena Zernicka-Goetz. 3D image segmentation supported by a point cloud. Discrete & Continuous Dynamical Systems - S, 2021, 14 (3) : 971-985. doi: 10.3934/dcdss.2020351 [4] Maika Goto, Kazunori Kuwana, Yasuhide Uegata, Shigetoshi Yazaki. A method how to determine parameters arising in a smoldering evolution equation by image segmentation for experiment's movies. Discrete & Continuous Dynamical Systems - S, 2021, 14 (3) : 881-891. doi: 10.3934/dcdss.2020233 [5] Editorial Office. Retraction: Xiaohong Zhu, Zili Yang and Tabharit Zoubir, Research on the matching algorithm for heterologous image after deformation in the same scene. Discrete & Continuous Dynamical Systems - S, 2019, 12 (4&5) : 1281-1281. doi: 10.3934/dcdss.2019088 [6] Editorial Office. Retraction: Xiaohong Zhu, Lihe Zhou, Zili Yang and Joyati Debnath, A new text information extraction algorithm of video image under multimedia environment. Discrete & Continuous Dynamical Systems - S, 2019, 12 (4&5) : 1265-1265. doi: 10.3934/dcdss.2019087 [7] Predrag S. Stanimirović, Branislav Ivanov, Haifeng Ma, Dijana Mosić. A survey of gradient methods for solving nonlinear optimization. Electronic Research Archive, 2020, 28 (4) : 1573-1624. doi: 10.3934/era.2020115 [8] Mehdi Bastani, Davod Khojasteh Salkuyeh. On the GSOR iteration method for image restoration. Numerical Algebra, Control & Optimization, 2021, 11 (1) : 27-43. doi: 10.3934/naco.2020013 [9] Aihua Fan, Jörg Schmeling, Weixiao Shen. $L^\infty$-estimation of generalized Thue-Morse trigonometric polynomials and ergodic maximization. Discrete & Continuous Dynamical Systems - A, 2021, 41 (1) : 297-327. doi: 10.3934/dcds.2020363 [10] Hassan Mohammad. A diagonal PRP-type projection method for convex constrained nonlinear monotone equations. Journal of Industrial & Management Optimization, 2021, 17 (1) : 101-116. doi: 10.3934/jimo.2019101 [11] Wolfgang Riedl, Robert Baier, Matthias Gerdts. Optimization-based subdivision algorithm for reachable sets. Journal of Computational Dynamics, 2021, 8 (1) : 99-130. doi: 10.3934/jcd.2021005 [12] Jie Zhang, Yuping Duan, Yue Lu, Michael K. Ng, Huibin Chang. Bilinear constraint based ADMM for mixed Poisson-Gaussian noise removal. Inverse Problems & Imaging, , () : -. doi: 10.3934/ipi.2020071 [13] Xi Zhao, Teng Niu. Impacts of horizontal mergers on dual-channel supply chain. Journal of Industrial & Management Optimization, 2020  doi: 10.3934/jimo.2020173 [14] Huyuan Chen, Dong Ye, Feng Zhou. On gaussian curvature equation in $\mathbb{R}^2$ with prescribed nonpositive curvature. Discrete & Continuous Dynamical Systems - A, 2020, 40 (6) : 3201-3214. doi: 10.3934/dcds.2020125 [15] Ferenc Weisz. Dual spaces of mixed-norm martingale hardy spaces. Communications on Pure & Applied Analysis, 2021, 20 (2) : 681-695. doi: 10.3934/cpaa.2020285 [16] Manxue You, Shengjie Li. Perturbation of Image and conjugate duality for vector optimization. Journal of Industrial & Management Optimization, 2020  doi: 10.3934/jimo.2020176 [17] Martin Heida, Stefan Neukamm, Mario Varga. Stochastic homogenization of $\Lambda$-convex gradient flows. Discrete & Continuous Dynamical Systems - S, 2021, 14 (1) : 427-453. doi: 10.3934/dcdss.2020328 [18] Gabrielle Nornberg, Delia Schiera, Boyan Sirakov. A priori estimates and multiplicity for systems of elliptic PDE with natural gradient growth. Discrete & Continuous Dynamical Systems - A, 2020, 40 (6) : 3857-3881. doi: 10.3934/dcds.2020128 [19] Petr Pauš, Shigetoshi Yazaki. Segmentation of color images using mean curvature flow and parametric curves. Discrete & Continuous Dynamical Systems - S, 2021, 14 (3) : 1123-1132. doi: 10.3934/dcdss.2020389 [20] Kateřina Škardová, Tomáš Oberhuber, Jaroslav Tintěra, Radomír Chabiniok. Signed-distance function based non-rigid registration of image series with varying image intensity. Discrete & Continuous Dynamical Systems - S, 2021, 14 (3) : 1145-1160. doi: 10.3934/dcdss.2020386

2019 Impact Factor: 1.373