Texture Segmentation of Jacquard Fabric Image Based on Multiresolution Markov Random Field

Abstract:

In order to develop an automated segmentation system for jacquard fabric images, a new approach based on MRMRF algorithm with variable weighing parameter is proposed in this paper. Firstly the variable weighting parameter different to the one in traditional MRMRF is described, which can provide a more accurate vector. The next step is MAP estimation and the model for texture segmentation. During this iterative process the initial value is big enough to learn more accurate parameters of feature energy. With the iterative number going on, the value will decrease and stop decreasing when the iterative number comes to some degree. Lastly the experiment results show that the new approach works better than the traditional method with constant weighing parameter.

• Periodical: Applied Mechanics and Materials (Volumes 101-102)
• Pages: 496-499
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.101-102.496
• Online since: September 2011
• Authors: You Cheng Tong, Yang Zhang, Jun Zhou Yao
• Keywords: Jacquard Fabric, Multiresolution Markov Random Field (MRMRF), Texture Segmentation

References

[1] S. Jiang. Journal of Donghua University(Eng. Ed. ), Vol. 3 (2008) No. 6, pp.718-724.

[2] H.L. Tong. Journal of Donghua University(Eng. Ed. ), Vol. 6 (2010) No. 2, pp.246-248.

[3] A.G. Miguel. Image and Vision Computing, Vol. 6 (2010) No. 25, pp.1091-1106.

[4] Y. Gong, N. Shu and J.L. Li. Geo-spatial Information Scienc, Vol. 8 (2010) No. 13, pp.16-23.

[5] G. Cehux, F. Forbes and N. Peyrard. Pattern Recognition, Vol. 1 (2003) No. 36, pp.131-144.