Unsupervised abdomen CT image segmentation using variable weight MRF in spatial and wavelet domain

Aiming at the segmentation of liver image with fuzzy edge, a new algorithm based on Markov Random Field in spatial and wavelet domains is proposed. Firstly, a lifting wavelet transform is adopted to represent an original image in different resolutions; Secondly, attain the low frequency subimage and execute an initial segmentation and a multi-level segmentation; Lastly, the spatial domain transform is applied on the segmentation result of the wavelet domain to revise the segmentation results and to get an accurate outcome. The algorithms of the initial and multi-level segmentation in wavelet domain are K-means improved by SAPSO and MAP/ICM. Experimental results show that the proposed algorithm has a good robustness.