模拟人眼视觉特性的高性能矢量多边形叠加分析算法研究

项目名称： 模拟人眼视觉特性的高性能矢量多边形叠加分析算法研究

项目编号： No.41501425

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 天文学、地球科学

项目作者： 范俊甫

作者单位： 山东理工大学

项目金额： 20万元

中文摘要： 空间数据快速累积使传统的矢量多边形叠加分析算法在处理包含大量顶点的几何对象时面临性能瓶颈，基于栅格化思想的叠加分析算法又难以克服误差难题，而结合并行计算技术的人眼视觉特性仿生模拟为解决上述问题提供了新的思路。本课题针对基于人眼视觉特性仿生的多边形叠加分析问题的高性能求解理论方法开展研究，拟：（1）研究人眼视觉侧抑制特性的相关性质与理论描述，建立面向矢量数据叠加分析的视觉侧抑制特性仿生计算模型；（2）基于视觉特性仿生计算模型设计GPU并行加速的平面非线性渐进式交点快速定位与精确解算算法，研发兼具高精度和高效率双重特征的多边形裁剪算法；（3）设计研发多并行计算环境/多并行计算粒度耦合驱动的高性能视觉仿生矢量多边形叠加分析算法集合。本研究以视觉仿生为切入点，可望进一步丰富矢量数据空间分析问题研究的理论与方法体系，并取得具有一定创新性的研究成果，是研究矢量多边形叠加分析问题的一次有益探索。

中文关键词： 矢量数据分析；多边形叠加；侧抑制特性；人眼视觉仿生；并行计算

英文摘要： As one of the basic spatial analysis functions in GIS, vector polygon overlay analysis is widely used in the fields of spatial data integration and update, spatial modeling and analysis, and information mining. Because of the rapid accumulation of spatial data, the traditional planar-scanning based vector polygon overlay analysis algorithms are facing performance bottlenecks when handling geometries with plenty vertices, and the rasterization based polygon overlay analysis algorithms can not prevent errors. However, the simulation of the visual lateral inhibition characteristic of human vision system combined with parallel computing technology provides a new approach to solving the problems. The research project focuses on the theories and methods of the high-performance solutions of polygon overlapping problems based on the simulation of human visual system. This research proposes the following three contents: (1) conduct research on the characteristics and theoretical description of human visual lateral inhibition property, build the visual lateral inhibition property based bionic computing model for vector spatial data analysis; (2) design the nonlinear gradual fast locating and coordinates precise calculating algorithms of polygon boundary intersection points in planar space, develop new polygon clipping algorithm with both characteristics of high precision and high efficiency; (3) based on the new polygon clipping algorithm, design and develop human vision bionic-based high performance polygon overlay analysis algorithm system which is coupling driven by multiple parallel computing environments and multiple parallel computing granularities. This research is a useful attempt to solve the vector polygon overlay analysis problem by taking the human vision bionics as a starting point. This research aims not only to further enrich the theoretical and methodology system of the research of vector data spatial analysis problem, but also to gain a breakthrough in the topic of performance promotion of polygon overlay analysis algorithm from the perspective of bionics.

英文关键词： vector data analysis;polygon overlay;lateral inhibition characteristic;human vision bionics;parallel computing