中文摘要：

面目标的聚集模式识别是空间聚类研究的重要方向之一,但因多边形几何信息和空间障碍阻隔的双重约束,目标的位置相似性难以快速而准确地计算。扩展点目标多尺度聚类方法,通过构建面目标的强度函数计算目标与邻近目标的位置聚集程度,提出了有效作用于双重约束下的面目标位置聚类法,并以判断相邻尺度下同一面目标类的强度函数阈值相等作为算法的收敛条件。经试验分析与比较发现,算法无须自定义参数,能够识别密度不均、任意形状分布,以及＂桥＂链接的面目标集群,同时能够准确判断障碍约束对面目标簇的阻隔和划分。

英文摘要：

It is a vital research direction for spatial clustering to recognize polygon cluster,but due to the dual-constrains by geometric information of polygons and obstacles,the position similarity of polygon is difficult to calculate accurately and quickly.A polygon clustering algorithm under dual-constrains is proposed by extending the algorithm of multi-scale spatial clustering,and constructing an intensity function to express position aggregation between object and its adjacent object.For further discuss,it takes the same thresholds of intensity function in adjacent scales as convergence condition.Simulated polygons and real data are chosen to perform clustering in experiments to verify the validity of our algorithm.Results show that without predefined parameters,this algorithm can identify variety polygon clusters with different densities,arbitrary shape,bridge and obstacle.