Research on segmentation algorithm for jacquard images using Mumford-Shah model
Feng Zhi · Journal of Zhejiang University(Engineering Science) · 2005
For solving the problem of low accuracy in segmentation of jacquard images under noisy environment, a numerical implementation algorithm was proposed by using the MumfordShah model. Based on Gammaconvergence and bounded variation functions theories, the minimization of the model was seen as a free discontinuity problem for the segmentation of noisy jacquard images. A discrete formulation of the model was defined on piecewise affine spaces of adaptive triangulation, and the model was approximated in the sense of Gammaconvergence by a sequence of the discrete formulations. An adaptive adjustment algorithm for the triangulation and the finite element mesh technique were enforced to characterize the essential contour structure of a jacquard pattern. The conjugate gradient method was utilized to find the absolute minimum of the discrete formulation. Experimental results show that the proposed algorithm is robust against noise, and can improve the integrity of the segmentation performance.