针对传统局部一致性方法的缺点,在研究全局一致性方法的基础上,提出一种对偶分布匹配(Dual Distribution Matching,即DDM)的图像分割算法。该算法首先将前景和背景的概率分布作为输入分布,构造出前景和背景的对偶匹配模型,该模型描述两...针对传统局部一致性方法的缺点,在研究全局一致性方法的基础上,提出一种对偶分布匹配(Dual Distribution Matching,即DDM)的图像分割算法。该算法首先将前景和背景的概率分布作为输入分布,构造出前景和背景的对偶匹配模型,该模型描述两个输入分布和分割结果的相似度,然后利用整幅图像的分布来确定模型的权重参数,从而求解能量函数ε(L)的全局最小化的真解L*,最后利用基于Bhattacharyya的图分割(Bhattacharyya Measure Graph Cut,BMGC)的辅助函数完成能量函数ε(L)的优化,不断更新辅助标记La,Lb收敛于真实标记L*。实验表明在输入分布不够精确的情况下,该算法具有较好的准确性和稳定性。展开更多
The Lee weight enumerators and the complete weight enumerators for the linear codes over ring R = F2 + u F2 + v F2 are defined and Gray map from R^nto F2^3n is constructed. By proving the fact that the Gray images o...The Lee weight enumerators and the complete weight enumerators for the linear codes over ring R = F2 + u F2 + v F2 are defined and Gray map from R^nto F2^3n is constructed. By proving the fact that the Gray images of the self-dual codes over R are the self-dual codes over F2, and based on the Mac Williams identities for the Hamming weight enumerators of linear codes over F2, the Mac Williams identities for Lee weight enumerators of linear codes over R are given. Further, by introducing a special variable t, the Mac Williams identities for the complete weight enumerators of linear codes over R are obtained. Finally, an example which illustrates the correctness and function of the two Mac Williams identities is provided.展开更多
基金supported by the Natural Science Foundation of Hubei Province under Grant No.D20144401the Natural Science Foundation of Hubei Polytechnic University under Grant Nos.12xjz14A,11yjz37B
文摘The Lee weight enumerators and the complete weight enumerators for the linear codes over ring R = F2 + u F2 + v F2 are defined and Gray map from R^nto F2^3n is constructed. By proving the fact that the Gray images of the self-dual codes over R are the self-dual codes over F2, and based on the Mac Williams identities for the Hamming weight enumerators of linear codes over F2, the Mac Williams identities for Lee weight enumerators of linear codes over R are given. Further, by introducing a special variable t, the Mac Williams identities for the complete weight enumerators of linear codes over R are obtained. Finally, an example which illustrates the correctness and function of the two Mac Williams identities is provided.