Automatic pattern segmentation of jacquard images plays an important role in jacquard pattern analysis. This paper dealt with the problem of low accuracy in segmentation of jacquard images under noisy environment. The phase field model was introduced to extract specific pattern structures within a jacquard image. A novel adaptive mesh minimization algorithm for numerical solving of the phase field model was also proposed. In this algorithm, the phase field model was discretized on finite element spaces of adaptive triangulation. Then, a mesh adjustment procedure for the adaptive triangulation was deployed to characterize the essential contour structure of a jacquard pattern. Finally, a quasi-Newton algorithm was applied to find the minimum of the discrete phase field model. Experimental results on synthetic and jacquard images show that the proposed algorithm is feasible and outperforms other algorithms.
