色彩混合原理支持下的空间同位模式可视化挖掘

空间同位模式挖掘是空间数据挖掘研究的重要内容，在城市设施规划与配置、生态共生现象发现、犯罪事件依赖关系探测等应用上可发挥重要作用。本研究一改传统方法主要依赖统计分析和数据挖掘的方式，将可视化的形象思维引入，建立空间同位模式可视化挖掘方法，充分展现可视化“视物致知”的功效，以期对经典空间数据挖掘方法进行有效的补充。本申请项目将以空间同位模式视觉模型的"构建—解译—优化"为研究主线，首先，构建基于色彩混合原理的视觉模型以模拟和表达空间同位模式；然后，对视觉模型机理进行深层次诠释和解译；最后，顾及视觉感知对模型进行优化。本研究将突破传统可视化主要用于研究结果符号化表达的局限，将其提升为一种视觉思维方法，可通过形象的方式直接探求、揭示和模拟空间现象的地学规律；并有望为空间同位模式挖掘提供新的思路和方法途径，提升同位模式挖掘的实际应用潜力。

Spatial co-location mining, which is one of the major spatial data-mining tasks, has wide applications in domains such as ecology, biology, business, epidemiology, public safety, and transportation. For instance, co-location mining can be used to judge the planning reasonability of facilities, reveal symbiotic species and detect dependence of crimes. Different from traditional statistical or data mining methods, visualization that based on visual thinking is introduced, showing the function that "begins in visual representation and ends in knowledge". This proposal aims to develop a visualization method to discover spatial co-location patterns. The visualization method can be an effective supplement to classical spatial data mining methods. A framework consisting of visualization model construction, interpretation and refinement is conceived. A visualization model is constructed supported by the color mixing technique. The mechanism of the visualization model is analyzed and interpreted. The visualization model is refined considering the visual perception of human. This study will break through the limitation that visualization mainly undertakes the function of results representation, and promote visualization up to a visual thinking approach to discover geographic knowledge. This proposal will provide new ideas and methods for co-location mining, and improve the potential of co-location mining applications.