A segmentation algorithm for jacquard images based on mumford-shah model
Zhilin Feng, Jianwei Yin, Gang Chen, Yang Liu, Jinxiang Dong · 2004
Automatic pattern segmentation of jacquard images is a challenging task due to the complexity of the images. Active contour models have become popular for finding the contours of a pattern with a complex shape. However, these models have many limitations on the pattern segmentation of jacquard images in the presence of noise. In this paper, a robust algorithm based on the Mumford-Shah model is proposed for the segmentation of noisy jacquard images. We discretize the Mumford-Shah model on piecewise lin-ear finite element spaces to yield greater stability and higher accuracy. A novel iterative relaxation algo-rithm for the numerical solving of the discrete version of the Mumford-Shah model is presented. During each iteration, we first refine and reorganize an adaptive triangular mesh to characterize the essential contour structure of a pattern. Then, we apply the quasi-Newton algorithm to find the absolute minimum of the discrete version of the model at the current iteration. Experimental results on synthetic and jac-quard images have shown the effectiveness and robustness of the algorithm.