针对低阶马尔科夫随机场(Markov random field, MRF)模型难以有效表达自然图像中复杂的先验知识而造成误分割问题,提出一种基于多节点拓扑重叠测度高阶MRF模型(Higher-order MRF model with multi-node topological overlap measure, MT...针对低阶马尔科夫随机场(Markov random field, MRF)模型难以有效表达自然图像中复杂的先验知识而造成误分割问题,提出一种基于多节点拓扑重叠测度高阶MRF模型(Higher-order MRF model with multi-node topological overlap measure, MTOM-HMRF)的图像分割方法.首先,为描述图像局部区域内多像素蕴含的复杂空间拓扑结构信息,利用多节点拓扑重叠测度建立图像局部区域的高阶先验模型;其次,利用较大的局部区域包含更多的标签节点信息能力,基于Pairwise MRF模型建立基于局部区域的部分二阶Potts先验模型,提高分割模型的抗噪能力;再次,为有效描述观察图像场与其标签场的似然特征分布,研究利用局部区域内邻接像素的Hamming距离引入图像局部空间相关性,建立局部空间一致性约束的高斯混合分布;最后,基于MRF框架建立用于图像分割的多节点拓扑重叠测度高阶MRF模型,采用Gibbs采样算法对提出模型进行优化.实验结果表明,提出模型不仅能有效抵抗图像强噪声和复杂的纹理突变干扰,鲁棒性更好,而且具有更准确的图像分割结果.展开更多
Overlap fermions and improved 5Li cooling are compared, which exhibits that these two filter methods can produce similar structures. Then the overlap fermions and improved 5Li cooling are combined to calculate the top...Overlap fermions and improved 5Li cooling are compared, which exhibits that these two filter methods can produce similar structures. Then the overlap fermions and improved 5Li cooling are combined to calculate the topological charge and identify the structure of the vacuum. The results are consistent with instanton liquid model .展开更多
This paper presents some results about bounds for coincidence indices of Nielsen coincidence classes for maps between nonorientable surfaces. Denoting by Kn the nonorientable surface constructed by a connected sum of ...This paper presents some results about bounds for coincidence indices of Nielsen coincidence classes for maps between nonorientable surfaces. Denoting by Kn the nonorientable surface constructed by a connected sum of n torus with a Klein bottle, the author proves: (i) for pairs of maps between two Klein bottles or for pairs of maps from a Klein bottle to a surface Kn the coincidence class index is bounded. (ii) for pairs of maps from Kn to the Klein bottle the coincidence class index is unbounded. Other boundedness results are given for more technical conditions, including one for self maps.展开更多
文摘针对低阶马尔科夫随机场(Markov random field, MRF)模型难以有效表达自然图像中复杂的先验知识而造成误分割问题,提出一种基于多节点拓扑重叠测度高阶MRF模型(Higher-order MRF model with multi-node topological overlap measure, MTOM-HMRF)的图像分割方法.首先,为描述图像局部区域内多像素蕴含的复杂空间拓扑结构信息,利用多节点拓扑重叠测度建立图像局部区域的高阶先验模型;其次,利用较大的局部区域包含更多的标签节点信息能力,基于Pairwise MRF模型建立基于局部区域的部分二阶Potts先验模型,提高分割模型的抗噪能力;再次,为有效描述观察图像场与其标签场的似然特征分布,研究利用局部区域内邻接像素的Hamming距离引入图像局部空间相关性,建立局部空间一致性约束的高斯混合分布;最后,基于MRF框架建立用于图像分割的多节点拓扑重叠测度高阶MRF模型,采用Gibbs采样算法对提出模型进行优化.实验结果表明,提出模型不仅能有效抵抗图像强噪声和复杂的纹理突变干扰,鲁棒性更好,而且具有更准确的图像分割结果.
基金The project supported in part by National Natural Science Foundation of China under Grant Nos. 90103018 and 90503011
文摘Overlap fermions and improved 5Li cooling are compared, which exhibits that these two filter methods can produce similar structures. Then the overlap fermions and improved 5Li cooling are combined to calculate the topological charge and identify the structure of the vacuum. The results are consistent with instanton liquid model .
文摘This paper presents some results about bounds for coincidence indices of Nielsen coincidence classes for maps between nonorientable surfaces. Denoting by Kn the nonorientable surface constructed by a connected sum of n torus with a Klein bottle, the author proves: (i) for pairs of maps between two Klein bottles or for pairs of maps from a Klein bottle to a surface Kn the coincidence class index is bounded. (ii) for pairs of maps from Kn to the Klein bottle the coincidence class index is unbounded. Other boundedness results are given for more technical conditions, including one for self maps.
