基于动态规划原理的半自动线特征提取

提出了一种针对遥感图像基于树结构实现的动态规划半自动线特征提取算法。在线模型的约束下 ,线特征的搜索过程很容易描述为一个多阶段决策的优化计算问题。文中提出了一种线特征模型 ,使之可以同样适合于遥感图像中的明、暗线特征的提取。并采用一种双向预测技术 ,使之也可适用于较大曲率线特征。最后利用树结构和动态剪枝技术实现整个线的半自动提取过程。实验表明了算法的有效性。

Dynamic Programming Based Semi-auto Linear Feature Extraction

A tree structure and dynamic programming theory based semi auto linear feature extraction method in remotely sensed imagery is proposed Under the constriction of line model,the line searching process can be described as a multiple phase decision optimization problem A generalized line model is provided,which can be applied to both salient and non salient line With a certain double direction predication technique,the model is suit for the sharp lines as well A tree structure accompanied with dynamic branch cutting is used to implemente the proposed method in the end and the validity of the algorithm is proved by the experiments