关键词: 相位展开, 图割, 模糊质量图, 标记

Abstract: Phase unwrapping is one of the most key steps in magnetic resonance imaging（MRI） applications in which accurate phase is required.The main difficulty in phase unwrapping stems from the noise and（or） abrupt variation in the true phase data.To address this problem，we present a method based on Markov Random Field & Maximum A Posterioi（MRF-MAP）.We regard phase unwrapping as labeling problem in computer vision for the first time.Considering of the characteristic of MRI phase data，we design a fuzzy quality map and then obtain an energy function which is suitable to be minimize by an efficient algorithm based on graph cuts.

Key words: phase unwrapping, graph cuts, fuzzy quality map, labeling