基于内容图像检索中针对显著方向的有向滤波器

目前有向滤波器主要适用于广义图像域,而不适用于狭义图像域.针对该问题,引入统计方法和模糊理论,在每个方向对狭义图像域子集进行聚簇分析,根据分析结果确定隶属度,把隶属度作为该方向纹理特征向量的权重,并且根据隶属度确定k显著方向.从k显著方向获取纹理特征组成纹理特征向量,利用欧几里德距离确定纹理特征向量间的相似性.实验表明,由π,π2,0,32π方向计算的纹理特征向量在匹配精度上优于从0,π4,π2以及34π计算的,而且该有向滤波器适用于狭义图像域·

面向数据流的加权聚类及演化分析研究

为解决无限数据流在有限内存空间中的聚类分析问题,本文提出了一种加权聚类及演化分析框架。为简要地描述此框架,给出了聚类、聚类簇的概念及其数据结构定义,接着对聚类、聚类簇的加法运算和差运算给出了清晰的描述和相应的实现算法。本框架与CluStream框架有较大的差别,这里采用聚类簇的加法运算来实现更大时间跨度内的聚类簇融合,采用聚类簇的差运算来进行聚类簇的演化分析。最后通过第一个例子来说明本框架是如何对数据流进行加权聚类及演化分析的,采用第二个例子来验证为实现本框架所需的十五个算法的正确性及有效性。

一种无线网络优化中挖掘LTE小区场景突变的方法研究

研究了一种检测无线网络优化中挖掘LTE小区场景突变的方法,通过筛选过滤出场景突变的LTE小区数据集,分析确认是否由于某些性能指标或者参数原因而导致的小区场景突变,最终挖掘得到发生场景突变的LTE小区。

A Method for Directly Counting and Quantitatively Comparing Aggregated Structures during Cluster Formation

Assembling of a few particles into a cluster commonly occurs in many systems.However,it is still challenging to precisely control particle assembling,due to the various amorphous structures induced by thermal fluctuations during cluster formation.Although these structures may have very different degrees of aggregation,a quantitative method is lacking to describe them,and how these structures evolve remains unclear.Therefore a significant step towards precise control of particle self-assembly is to describe and analyze various aggregation structures during cluster formation quantitatively.In this work,we are motivated to propose a method to directly count and quantitatively compare different aggregated structures.We also present several case studies to evaluate how the aggregated structures during cluster formation are affected by external controlling factors,e.g.,different interaction ranges,interaction strengths,or anisotropy of attraction.