The Image Segmentation System Based on Gauss Process

Abstract:

This paper gave an example for the design of automatic image segmentation system by using deep staining of blood cell image. The paper also described how to auto-locate the target position, and how to collect training samples with large entropy further. The spatial information of target object also contained valid information, so this paper put forward to use the relative distance between the inner points and the centre of a circle as a feature of a training sample to work together with the RGB features. And for the segmentation image can be applied to the later medical diagnosis conveniently, the Gauss process classifier had been used in medical image segmentation firstly because of its clear probabilistic interpretation. Compared with SVM, GP is better in this system.

• Periodical: Applied Mechanics and Materials (Volume 421)
• Pages: 523-527
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.421.523
• Online since: September 2013
• Authors: Lei Ling, Pan Chen, Li Ping
• Keywords: Cell, Entropy, Gauss Process, Segmentation

References

[1] Yiguang Hong, Daizhan Chen. Analysis and control of nonlinear systems. Beijing: Science press, (2005).

[2] Yimin Sun, Lei Guo. Global asymptotic controllability on planar affine nonlinear system. China scientific (E), 35(8): 830-839, (2005).

[3] WILLIAMS C K I, RASMUSSEN C E. Gaussian processes for regression. Cambridge, USA: MIT Press, (1996).

[4] Kadir T, Brady M. Saliency, Scale and Image Description[J]. International Journal of Computer Vision, 2001, 45(2): 83-105.

[5] Chen Pan, Xiangguo Yan, and Chongxun Zheng. Hard Margin SVM for Biomedical Image Segmentation[J]. Lecture Notes in Computer Science. SpringerLink. 2005, Vol 3498, 754-759.

DOI: 10.1007/11427469_120