为了应对全球70%的未注册土地权的挑战,对地籍测绘方法的需求不断增长。由于传统的现场实地测量既耗时又耗费人力,因此土地管理部门一直提倡基于遥感的地籍测绘,但基于遥感影像的自动划界的准确性仍然是一项重大挑战。在这项研究中,使...为了应对全球70%的未注册土地权的挑战,对地籍测绘方法的需求不断增长。由于传统的现场实地测量既耗时又耗费人力,因此土地管理部门一直提倡基于遥感的地籍测绘,但基于遥感影像的自动划界的准确性仍然是一项重大挑战。在这项研究中,使用无人机获得的图像来探索深度全卷积网络(Fully Convolu-tional Networks,FCN)在城市和城郊地区进行地籍边界提取的能力。在甘肃天水的两个地点使用其他最先进的技术来测试FCN、多分辨率分割(Multi-Resolution Segmentation,MRS)和全局化边界概率(Globalized Probability of Boundary,gPb)算法的性能。实验结果表明:FCN在两个研究领域的表现均优于MRS和gPb,精度平均为0.79,召回率为0.37,F评分为0.50。总之,FCN能够有效地提取地籍边界,尤其是在大量地籍边界可见的情况下。这种自动化方法可以最大限度地减少手动数字化并减少实地工作,从而促进当前的地籍测绘和更新做法。展开更多
传统随机游走图像分割需要多次交互设置种子点以获得理想的分割结果。在视觉注意的基础上,提出了一种新的自动确定种子点的随机游走图像分割算法。首先对图像进行超像素分割,并生成概率边界图(PBM);然后基于Itti模型,通过视觉注意焦点...传统随机游走图像分割需要多次交互设置种子点以获得理想的分割结果。在视觉注意的基础上,提出了一种新的自动确定种子点的随机游走图像分割算法。首先对图像进行超像素分割,并生成概率边界图(PBM);然后基于Itti模型,通过视觉注意焦点的转移搜寻待分割的关键区域;为确定关键分割区域种子点,以当前注意焦点作为极点对概率边界图进行极坐标变换,在获得的极坐标概率边界图上建立关于焦点区域边界的能量函数,采用图论max-flow min-cut算法最小化能量函数检测焦点区域的最优边界,焦点区域边界内的超像素即为种子点;最后以超像素为节点构造图,在图上随机游走完成图像分割。在Berkeley Segmentation Data Set上的实验表明本文方法能有效分割复杂图像。展开更多
Consider a family of probability measures {vξ} on a bounded open region D C Rd with a smooth boundary and a positive parameter set {βξ}, all indexed by ξ∈δD. For any starting point inside D, we run a diffusion u...Consider a family of probability measures {vξ} on a bounded open region D C Rd with a smooth boundary and a positive parameter set {βξ}, all indexed by ξ∈δD. For any starting point inside D, we run a diffusion until it first exits D, at which time it stays at the exit point ξ for an independent exponential holding time with rate βξ and then leaves ξ by a jump into D according to the distribution ξ. Once the process jumps inside, it starts the diffusion afresh. The same evolution is repeated independently each time the process jumped into the domain. The resulting Markov process is called diffusion with holding and jumping boundary （DHJ）, which is not reversible due to the jumping. In this paper we provide a study of DHJ on its generator, stationary distribution and the speed of convergence.展开更多
文摘为了应对全球70%的未注册土地权的挑战,对地籍测绘方法的需求不断增长。由于传统的现场实地测量既耗时又耗费人力,因此土地管理部门一直提倡基于遥感的地籍测绘,但基于遥感影像的自动划界的准确性仍然是一项重大挑战。在这项研究中,使用无人机获得的图像来探索深度全卷积网络(Fully Convolu-tional Networks,FCN)在城市和城郊地区进行地籍边界提取的能力。在甘肃天水的两个地点使用其他最先进的技术来测试FCN、多分辨率分割(Multi-Resolution Segmentation,MRS)和全局化边界概率(Globalized Probability of Boundary,gPb)算法的性能。实验结果表明:FCN在两个研究领域的表现均优于MRS和gPb,精度平均为0.79,召回率为0.37,F评分为0.50。总之,FCN能够有效地提取地籍边界,尤其是在大量地籍边界可见的情况下。这种自动化方法可以最大限度地减少手动数字化并减少实地工作,从而促进当前的地籍测绘和更新做法。
文摘传统随机游走图像分割需要多次交互设置种子点以获得理想的分割结果。在视觉注意的基础上,提出了一种新的自动确定种子点的随机游走图像分割算法。首先对图像进行超像素分割,并生成概率边界图(PBM);然后基于Itti模型,通过视觉注意焦点的转移搜寻待分割的关键区域;为确定关键分割区域种子点,以当前注意焦点作为极点对概率边界图进行极坐标变换,在获得的极坐标概率边界图上建立关于焦点区域边界的能量函数,采用图论max-flow min-cut算法最小化能量函数检测焦点区域的最优边界,焦点区域边界内的超像素即为种子点;最后以超像素为节点构造图,在图上随机游走完成图像分割。在Berkeley Segmentation Data Set上的实验表明本文方法能有效分割复杂图像。
基金supported by National Natural Science Foundation of China(Grant No.11101433)the Fundamental Research Funds for the Central South University(Grant No.2011QNZT105)+1 种基金Doctorial Dissertation Program of Hunan Province(Grant No.YB2011B009)US National Science Foundation (Grant Nos.AMC-SS-0706713,DMS-0805929,NSFC-6398100 and CAS-2008DP173182)
文摘Consider a family of probability measures {vξ} on a bounded open region D C Rd with a smooth boundary and a positive parameter set {βξ}, all indexed by ξ∈δD. For any starting point inside D, we run a diffusion until it first exits D, at which time it stays at the exit point ξ for an independent exponential holding time with rate βξ and then leaves ξ by a jump into D according to the distribution ξ. Once the process jumps inside, it starts the diffusion afresh. The same evolution is repeated independently each time the process jumped into the domain. The resulting Markov process is called diffusion with holding and jumping boundary （DHJ）, which is not reversible due to the jumping. In this paper we provide a study of DHJ on its generator, stationary distribution and the speed of convergence.
