空间聚类最优化分析

Spatial clustering is widely used in many fields such as WSN （Wireless Sensor Networks）, web clustering, remote sensing and so on for discovery groups and to identify interesting distributions in the underlying database. By discussing the relationships between the optimal clustering and the initial seeds, a clustering validity index and the principle of seeking initial seeds were proposed, and on this principle we recommend an initial seed-seeking strategy： SSPG （Single-Shortest-Path Graph）. With SSPG strategy used in clustering algorithms, we find that the result of clustering is optimized with more probability. At the end of the paper, according to the combinational theory of optimization, a method is proposed to obtain optimal reference k value of cluster number, and is proven to be efficient.     

基于SSPG的空间聚类初始种子选择方法

提出了初始种子选取原则及其基于这个原则的“单连通最短路径图”的初始种子选择方法,能自动发现数据集种密度相对较大的区域,并根据预先设定的聚类数目自动获寥最优的初始种子集合。实验结果证明了该方法具有较高的正确性,有效提高了空间聚类效率。     

一种利用像素邻域信息的模糊聚类图像分割算法

本文提出一种结合空间信息的模糊C均值聚类图像分割算法。该方法是利用每个像素的邻域像素的隶属度来修正FCM算法的隶属度函数,从而引入图像的空间信息,对隶属函数做了改进;依据平方误差和最小准则,从而确定模糊分类矩阵及聚类中心;并依据最大隶属度原则,划分图像像素的类别归属。实验结果表明,该方法能快速有效地分割图像,并且具有较强的抗噪能力。     

高光谱影像的改进K-均值监督式聚类分析方法

针对K-均值聚类存在的初始聚类中心不稳定、聚类数目难以确定的问题,提出利用正交投影散度(OPD)优化K-均值算法的初始聚类中心,设计了RD指标函数用于估计聚类数目k。将所提出的算法应用于高光谱影像特征提取与端元提取分析,实验结果表明,所提出算法的性能高于已有的类似算法。     

基于均值-标准差的K均值初始聚类中心选取算法

遥感图像分类是遥感图像处理中长期存在的一个难题，针对不同的传感器图像，不同的应用需求，选择合适的分类算法非常重要。在分类中不仅要考虑分类的精度，而且要考虑分类效率。本文研究了K均值算法的初始聚类中心的选择对算法本身聚类精度及效率的影响，提出了一种高效高精度的初始聚类中心选取方案，实验结果表明。利用该算法进行地表分类，效率比ENVI的K-Means（K均值）模块高。     

空间知识挖掘的自然面群聚集度聚类方法

空间聚类是挖掘空间知识的重要手段之一.针对现有方法难以处理几何、分布特征差异大的面群聚类问题,本文提出了一种面要素分布密度的描述参数—聚集度,并设计了一种自然面群聚类方法.首先,分析了面要素分布密度的影响因子,定义了聚集度的概念,设计其计算方法并验证其有效性及优势;然后,基于聚集度和边界最短距离建立相邻面从属关系,识别...     

自适应差分进化的遥感影像自动模糊聚类方法

遥感影像模糊聚类方法可以在无需样本分布信息的情况下获取比硬聚类方法更高的分类精度,但其仍依赖先验知识来确定影像地物的类别数。本文提出了一种基于自适应差分进化的遥感影像自动模糊聚类方法,该方法利用差分进化搜索速度快、计算简单、稳定性高的优点,以Xie-Beni指数为优化的适应度函数,在无需先验类别信息的情况下自动判定图像的类别数,并结合局部搜索算子对遥感影像进行最优化聚类。通过模拟影像以及两幅真实遥感图像的分类实验表明,本文方法不仅可以正确地自动获取地物类别数,而且能够获得比K均值、ISODATA以及模糊K均值方法更高的分类精度。     

基于马尔科夫随机场与模糊C均值的遥感影像分类

模糊C均值聚类是一种经典的非监督聚类模型,成功地应用于遥感影像分类。但是该方法对初始值敏感,容易陷入局部最优解;同时聚类时仅考虑光谱信息,忽略了空间信息。本文提出了一种新的基于马尔科夫随机场的模糊C均值聚类方法,该方法把马尔科夫随机场和模糊C均值结合在一起。初始值依据第一主成分的密度函数确定,这样克服了对初始值的依赖性,又在聚类的时候考虑了空间信息。通过实例数据验证,所提出的方法分类精度优于传统的模糊C均值模型。     

基于自组织神经网络的空间点群聚类及其应用分析

探讨了采用自组织神经网络进行离散空间点群聚类的原理、方法及应用分析,提出了一种兼顾几何距离和属性特征的广义Euclid距离,并将其作为聚类统计量.并以实例验证了采用自组织空间聚类进行空间点群的数据分类、异常数据检验、均质区域划分等是有效的.     

基于共享单车时空大数据的细粒度聚类

针对传统上单独采用K-means或DBSCAN等方法对共享单车位置数据聚类时造成的聚类结果与真实的聚类结构不符的问题,本文提出了一种基于共享单车时空大数据的细粒度聚类方法(FGCM).该方法通过DBSCAN进行初始聚类,并在此基础上采用GMM-EM算法进行细部聚类,以提取细粒度层级的热点区域.试验表明,该方法可根据密度...     

ParSymG: a parallel clustering approach for unsupervised classification of remotely sensed imagery

ABSTRACT

Symmetry is a common feature in the real world. It may be used to improve a classification by using the point symmetry-based distance as a measure of clustering. However, it is time consuming to calculate the point symmetry-based distance. Although an efficient parallel point symmetry-based K-means algorithm (ParSym) has been propsed to overcome this limitation, ParSym may get stuck in sub-optimal solutions due to the K-means technique it used. In this study, we proposed a novel parallel point symmetry-based genetic clustering (ParSymG) algorithm for unsupervised classification. The genetic algorithm was introduced to overcome the sub-optimization problem caused by inappropriate selection of initial centroids in ParSym. A message passing interface (MPI) was used to implement the distributed master–slave paradigm. To make the algorithm more time-efficient, a three-phase speedup strategy was adopted for population initialization, image partition, and kd-tree structure-based nearest neighbor searching. The advantages of ParSymG over existing ParSym and parallel K-means (PKM) alogithms were demonstrated through case studies using three different types of remotely sensed images. Results in speedup and time gain proved the excellent scalability of the ParSymG algorithm.     

A Clustering‐Based Approach to the Capacitated Facility Location Problem1

This research develops a clustering‐based location‐allocation method to the Capacitated Facility Location Problem (CFLP), which provides an approximate optimal solution to determine the location and coverage of a set of facilities to serve the demands of a large number of locations. The allocation is constrained by facility capacities – different facilities may have different capacities and the overall capacity may be inadequate to satisfy the total demands. This research transforms this special location‐allocation problem into a clustering model. The proposed approach has two parts: (1) the allocation of demands to facilities considering capacity constraints while minimizing the cost; and (2) the iterative optimization of facility locations using an adapted K‐means clustering method. The quality of a location‐allocation solution is measured using an objective function, which is the demand‐weighted distance from demand locations to their assigned facilities. The clustering‐based method is evaluated against an adapted Genetic Algorithm (GA) alternative, which integrates the allocation component as described above but uses GA operations to search for ‘optimal’ facility locations. Experiments and evaluations are carried out with various data sets (including both synthetic and real data).     

签到位置数据的密度峰值快速搜索与聚类方法

位置签到数据蕴含了城市居民活动变化。由于客户端位置候选问题,不同的签到行为以同一候选位置签到时会产生位置重复现象。针对现有密度聚类方法在签到数据聚类上存在的问题,以快速搜索和查找密度峰值聚类算法(CFSFDP)为基础,提出了签到位置数据的密度峰值快速搜索与聚类方法。首先,引入位置重复频率来表达签到位置重复,然后,对原始签到位置数据点统计位置重复频率并重新设计数据结构,以新的空间点要素为研究对象寻找密度峰值点;最后,构建了峰值点密度簇聚类算法,在点要素集聚类过程中考虑密度连通性来保证峰值密度簇的连续与完整。试验表明,所提出的聚类方法有效避免了重复度较高的离群位置对象选为峰值并聚类的情况,并具有良好的空间适应性。所提取的密度峰值点不仅可以用来表示热区的中心,还能够反映热区的集中趋势,进而可以帮助探索热区的动态变化情况。     

一种可见光影像稠密匹配点云的单栋建筑物提取方法

建筑物点云提取是城市快速三维建模的基础。针对城区中建筑物和树木空间距离较近导致建筑物点云误提取的问题,提出一种颜色约束的欧式聚类算法。该方法利用低空拍摄可见光影像进行三维重建、获取点云数据,在建立点云K邻域索引和表面估计的基础上,以曲率最小的点作为欧式聚类的种子点,将点云的RGB值转换成Lab颜色模型,对建筑物点云的聚类提取进行约束。实验表明,该方法可以有效地解决可见光影像匹配点云中建筑物提取时将树木误提取的问题。     

k均值聚类引导的遥感影像多尺度分割优化方法

针对不同尺度地物的分割需求,提出了一种k均值聚类引导的多尺度分割优化方法。首先对原始影像进行小尺度分割和k均值聚类,然后利用k均值聚类结果引导对象合并,在合并过程中利用Otsu阈值方法自动选择k均值聚类的影响因子,最终得到适应不同尺度地物的分割结果。以FNEA多尺度分割方法为例,利用模拟数据和真实的GeoEye-1影像数据进行相关试验,目视和定量评价表明本文方法能够得到适宜不同尺度地物的高质量分割结果。     

DCAD: a dual clustering algorithm for distributed spatial databases

Spatial objects have two types of attributes: geometrical attributes and non-geometrical attributes, which belong to two different attribute domains (geometrical and non-geometrical domains). Although geometrically scattered in a geometrical domain, spatial objects may be similar to each other in a non-geometrical domain. Most existing clustering algorithms group spatial datasets into different compact regions in a geometrical domain without considering the aspect of a non-geometrical domain. However, many application scenarios require clustering results in which a cluster has not only high proximity in a geometrical domain, but also high similarity in a non-geometrical domain. This means constraints are imposed on the clustering goal from both geometrical and non-geometrical domains simultaneously. Such a clustering problem is called dual clustering. As distributed clustering applications become more and more popular, it is necessary to tackle the dual clustering problem in distributed databases. The DCAD algorithm is proposed to solve this problem. DCAD consists of two levels of clustering: local clustering and global clustering. First, clustering is conducted at each local site with a local clustering algorithm, and the features of local clusters are extracted. Second, local features from each site are sent to a central site where global clustering is obtained based on those features. Experiments on both artificial and real spatial datasets show that DCAD is effective and efficient.     

K均值聚类算法的研究与并行化改进

K均值算法是一种常用的聚类分析方法,广泛应用于图像处理和机器学习等领域。但该算法具有较高的计算复杂度,导致了算法具有较大的局限性。为了提高算法的运行效率,本文在深入分析算法基本原理的基础上,利用CUDA架构提供的强大计算能力对该算法进行了并行化改进。实验结果表明,算法在取不同的聚类数时均取得了较高的加速比。     

DCAD： a Dual Clustering Algorithm for Distributed Spatial Databases

Spatial objects have two types of attributes: geometrical attributes and non-geometrical attributes, which belong to two different attribute domains (geometrical and non-geometrical domains). Although geometrically scattered in a geometrical domain, spatial objects may be similar to each other in a non-geometrical domain. Most existing clustering algorithms group spatial datasets into different compact regions in a geometrical domain without considering the aspect of a non-geometrical domain. However, many application scenarios require clustering results in which a cluster has not only high proximity in a geometrical domain, but also high similarity in a non-geometrical domain. This means constraints are imposed on the clustering goal from both geometrical and non-geometrical domains simultaneously. Such a clustering problem is called dual clustering. As distributed clustering applications become more and more popular, it is necessary to tackle the dual clustering problem in distributed databases. The DCAD algorithm is proposed to solve this problem. DCAD consists of two levels of clus- tering: local clustering and global clustering. First, clustering is conducted at each local site with a local clustering algorithm, and the features of local clusters are extracted. Second, local features from each site are sent to a central site where global clustering is obtained based on those features. Experiments on both artificial and real spatial datasets show that DCAD is effective and efficient.     

基于改进聚类分裂的动态R-树实现方法

在R*-树的构建过程中引入聚类技术能够有效地提高索引的性能,传统的k-means聚类算法对初始值非常敏感,聚类过程较为复杂。基于此,文中提出一种改进聚类分裂的动态R*-树实现方法,在节点分裂的过程中引进聚类技术,对R*-树的基本结构加以改进,从而获得动态的结构重组。实验表明,动态R*-树以略高的构建开销换取较高的查询效率,大幅度提高索引树的空间利用率,在批量数据动态加载和处理等方面具有较高的实用价值。     

复杂环境下高分二号遥感影像的城市地表水体提取

水体指数可以抑制背景噪声和提高地表水体的可分性,已经广泛用于地表水体提取。传统FCM聚类算法考虑了地物的不确定性,但没有顾及地物的邻域空间信息,对背景异质性比较敏感。针对传统FCM聚类算法的不足,提出一种可变邻域的区域FCM聚类算法。由于复杂环境下高分二号(GF-2)遥感影像的城市地表水体具有复杂异质背景和不确定性的特点,本文利用水体指数和区域FCM聚类算法的优点,提出一种整合水体指数和区域FCM的城市地表水体自动提取算法,该算法主要步骤包括:(1)去除影像阴影后计算归一化差分水体指数NDWI(Normalized Difference Water Index);(2)区域FCM聚类算法;(3)整合水体指数和区域FCM聚类的城市地表水体自动提取算法。最后采用两景GF-2高分辨率遥感影像(广州和武汉)进行实验,验证了该算法的有效性,并与经典地表水体提取算法进行对比分析。实验结果表明:该算法具有较高的水体提取精度,城市地表水体边界既具有较好的区域完整性又保持了局部细节,同时对城市地表水体复杂背景噪声具有较好的抑制作用,有效减少传统FCM聚类算法的"胡椒盐"现象。