Edge Multi-scale Markov Random Field Model Based Medical Image Segmentation in Wavelet Domain
The segmentation algorithms based on MRF often exist edge block effect, and have low operation efficiency by modeling the whole image. To solve the problems the image segmentation algorithm using edge multiscale domain hierarchical Markov model is presented. It views an edge as an observable data series, the image characteristic field is built on a series of edge extracted by wavelet transform, and the label field MRF model based on the edge is established to integrate the scale interaction in the model, then the image segmentation is obtained. The test images and medical images are experimented, and the results show that compared with the WMSRF algorithm, the proposed algorithm can not only distinguish effectively different regions, but also retain the edge information very well, and improve the efficiency. Both the visual effects and evaluation parameters illustrate the effectiveness of the proposed algorithm.
Keywordsmedical image segmentation MRF wavelet edge
Unable to display preview. Download preview PDF.
- 1.Qin, X.-J., Du, Y.-C., Zhang, S.-Q., et al.: Boundary Information Based C_V Model Method for Medical Image Segmentation. Journal of Chinese Computer Systems 32(5), 972–977 (2011) (in Chinese)Google Scholar
- 2.Sun, R.-M.: A New Region Growth Method for Medical Image Segment. Journal of Dalian Jiaotong University 31(2), 91–94 (2010) (in Chinese)Google Scholar
- 5.Tang, W., Zhang, C., Zhang, X., et al.: Medical Image Segmentation Based on Improved FCM. Journal of Computational Information System 8(2), 1–8 (2012)Google Scholar
- 11.Liu, G., Ma, G., Wang, L., et al.: Image modeling and segmentation in wavelet domain based on Markov Random Field——Matlab Environment. Science Press, Beijing (2010)Google Scholar
- 12.Zhang, Y.: Image Segmentation. Science Press, Beijing (2001)Google Scholar