犯罪轨迹时空变化建模与预警分析方法研究

Monitoring the trajectories changes of dangerous persons and detecting the unusual activities in time for security prediction are very important for Policing Technology in the situation of limited police forces and financial resources. Untill now, most analysis methods of criminal change focus on the whole area criminal distribution, so there is less study on modeling the spatio-temporal change and analysis methods of security prediction oriented single criminal object. Howerer, it can be found that the criminal trajectory has different spatio-temporal distribution pattern from the general GIS object, and the current spatio-temporal change model cannot meet the demand of trajectory classification and detection. also there is still not an effective method of safety prediction based on trajectory change analysis. So this study presented and studied the modeling the space-temporal change of crime trajectory, and analysis methods of security prediction. The main contents include: changes classification of the criminal trajectory; constructed trajectory change detection model considering the spatial and temporal boundary inconsistency; developed analysis methods of security prediction based on space-behavior relational degree; designed and implemented the prototype system for criminal trajectory changes monitoring and security prediction. This study will be great significance for finding the spatio-temporal regular pattern in criminal activities, and revealing the relationship between the crime activities and the trajectories. And also it will provide new clues for theoretical research of spatio-temporal changes and security prediction analysis.

对重点监控目标的犯罪轨迹进行变化监测，及时发现轨迹中的异常活动并做出安全预警，是实现科技强警的重要技术手段。迄今为止，犯罪时空分析大多从区域整体犯罪分布的角度，目前亟需发展面向犯罪个体目标的时空建模与预警分析方法。然而，犯罪轨迹与一般GIS时空目标有不同的时空表现形式与分布特征，现有时空变化建模方法不能满足犯罪轨迹分类描述与检测计算等需求，且缺乏有效的轨迹变化预警分析方法。本课题提出并研究犯罪轨迹时空建模与预警分析方法，主要研究内容包括：研究犯罪轨迹的变化分类；构建顾及时空边界不一致性的轨迹变化检测模型；发展基于空间行为关联度的预警分析方法；设计并实现犯罪轨迹时空变化监测与预警原型系统。本研究对发现犯罪活动的时空变化规律，揭示犯罪行为与时空轨迹间的关系具有重要意义，为时空变化理论研究及科学预警提供新线索。

背景：重点目标的犯罪轨迹变化监测是实现科技强警的重要技术手段。近年来对犯罪时空分析大多从区域整体犯罪分布的角度，而缺少面向犯罪个体目标的时空建模与预警分析方法。针对这一问题，本课题提出并研究犯罪轨迹时空建模与预警分析方法。主要研究内容包括：（1）研究犯罪轨迹的变化分类；（2）构建顾及时空边界不一致性的轨迹变化检测模型；（3）发展基于空间行为关联度的预警分析方法；（4）设计并实现犯罪轨迹时空变化监测与预警原型系统。重要结果：（1）语义变化是对几何变化的解释和描述；（2）时空边界的不一致性是造成变化识别中产生误差的根本原因之一；（3）异常活动轨迹可通过距离、时间统计得出。科学意义：可用于识别异常的犯罪活动，揭示犯罪行为与时空轨迹间的关系，为时空变化理论研究及科学预警提供新线索。