彩色地形图中地理要素提取与识别技术研究

大幅面彩色地形图中地理要素的提取与识别是地形图数字矢量化的基础，本课题针对目前地形图中地理要素提取与识别方面存在的难题，通过关键技术的探索与创新，分别在线划要素提取与分割、点状符号与文字注记提取与识别等方面进行深入研究。针对传统的图像分割算法在大幅面彩色地形图图像分割过程中存在的计算复杂度高和分割不精确等问题，以线划要素的同质区域为分割对象，研究基于区域特征和种子扩散的地形图中线划要素提取算法以及基于区域特征的线划要素分割算法。针对传统符号识别算法难以检测全部点状符号的问题，研究基于GHT的彩色地形图中点状符号直接识别算法。针对大幅面地形图中含有众多字符串和孤立字符的情况，通过分析地形图中文字注记在颜色、尺寸大小和方向等方面的特点，探索图模型和文字注记在地形图中的分布之间的关系，研究基于图模型的文字注记自动编组算法。这些关键技术的突破，对于大幅面彩色地形图高效数字化和矢量化，具有重要意义。

The extraction and recognition of the geographical elements in large color topographic map is the basis of the map vectorization. This project analyses the problems of the extraction and recognition of the geographical elements, and explores the new methods and ideas in the key problems. We research the problems about line segmentation and the identification of point symbols and texts.There are difficulties for traditional methods to segment these large maps because of the huge amount of computation and long computation time,in order to solve this problem, the methods based on the area features and seed spreading are used to seperate and segment the lines.In this method, the line elements are segmented, rather than pixels.In the research on the symbol recognition,the traditional methods cannot detect all the symbols in the maps, and a new method based on GHT is proposed to detect the symbols directly in topographic maps.A topographic map contains a large amount of strings and isolated character, by analyzing the characteristic of characters in color, size and direction, the distributing relaionship between graph model and character note is explored and the automatic text annotation grouping algorithm based on graph model is researched.