Given a road network G = （V, E）, where V（E） denotes the set of vertices（edges） in G, a set of points of interest P and a query point q residing in G, the reverse furthest neighbors （RFNR） query in road network...Given a road network G = （V, E）, where V（E） denotes the set of vertices（edges） in G, a set of points of interest P and a query point q residing in G, the reverse furthest neighbors （RFNR） query in road networks fetches a set of points p ∈ P that take q as their furthest neighbor compared with all points in P ∪ {q}. This is the monochromatic RFNR （MRFNR） query. Another interesting version of RFNR query is the bichromatic reverse furthest neighbor （BRFNR） query. Given two sets of points P and Q, and a query point q ∈ Q, a BRFNR query fetches a set of points p ∈ P that take q as their furthest neighbor compared with all points in Q. This paper presents efficient algorithms for both MRFNR and BRFNR queries, which utilize landmarks and partitioning-based techniques. Experiments on real datasets confirm the efficiency and scalability of proposed algorithms.展开更多
提出一种基于B样条概率密度函数(PDF,probability density function)估计的复杂纹理图像分类识别方法,主要包括图像纹理PDF特征学习、表征以及纹理类型识别3个步骤。在图像纹理PDF特征学习和表征中,引入各向异性高斯导数方向滤波器获得...提出一种基于B样条概率密度函数(PDF,probability density function)估计的复杂纹理图像分类识别方法,主要包括图像纹理PDF特征学习、表征以及纹理类型识别3个步骤。在图像纹理PDF特征学习和表征中,引入各向异性高斯导数方向滤波器获得纹理图像多尺度和多方向空间结构;然后基于预先固定的B样条基函数,将图像空间结构PDF估计转化为与基函数相对应的权值向量估计;之后采用最远邻聚类方法,获得图像空间纹理结构的PDF特征字典库;最后采用最近邻方法,获得各类纹理在特征字典库上的直方图分布表示。在纹理类型识别阶段,基于直方图距离测量结果实现纹理图像分类识别。在不同纹理图像数据库上进行了大量的验证性和对比性实验,实验结果表明所提方法的有效性和优越性。展开更多
基金This work was supported by the National Natural Science Foundation of China under Grant Nos. U1636210, 61472039, 61373156, 91438121, and 61672351, the National Basic Research 973 Program of China under Grant No. 2015CB352403, the National Key Research and Development Program of China under Grant Nos. 2016YFB0700502, 2016YFC0803000, and 2016YFB0502603, the Scientific Innovation Act of Science and Technology Commission of Shanghai Municipality under Grant No. 15JC1402400, and Microsoft Research Asia.
文摘Given a road network G = （V, E）, where V（E） denotes the set of vertices（edges） in G, a set of points of interest P and a query point q residing in G, the reverse furthest neighbors （RFNR） query in road networks fetches a set of points p ∈ P that take q as their furthest neighbor compared with all points in P ∪ {q}. This is the monochromatic RFNR （MRFNR） query. Another interesting version of RFNR query is the bichromatic reverse furthest neighbor （BRFNR） query. Given two sets of points P and Q, and a query point q ∈ Q, a BRFNR query fetches a set of points p ∈ P that take q as their furthest neighbor compared with all points in Q. This paper presents efficient algorithms for both MRFNR and BRFNR queries, which utilize landmarks and partitioning-based techniques. Experiments on real datasets confirm the efficiency and scalability of proposed algorithms.
文摘提出一种基于B样条概率密度函数(PDF,probability density function)估计的复杂纹理图像分类识别方法,主要包括图像纹理PDF特征学习、表征以及纹理类型识别3个步骤。在图像纹理PDF特征学习和表征中,引入各向异性高斯导数方向滤波器获得纹理图像多尺度和多方向空间结构;然后基于预先固定的B样条基函数,将图像空间结构PDF估计转化为与基函数相对应的权值向量估计;之后采用最远邻聚类方法,获得图像空间纹理结构的PDF特征字典库;最后采用最近邻方法,获得各类纹理在特征字典库上的直方图分布表示。在纹理类型识别阶段,基于直方图距离测量结果实现纹理图像分类识别。在不同纹理图像数据库上进行了大量的验证性和对比性实验,实验结果表明所提方法的有效性和优越性。
