基于形状模板匹配的多边形查询

Polygonal Inquiry Based on Shape Template Matching

  • 摘要: 对矢量目标如建筑物、注记等形状的识别通常是基于栅格数据的,研究的基点是像素,其不足之处是准确率不高,运算量过大。对此,提出了一种新的研究思路,用矢量取代栅格(像素)作为研究的基点,识别算法以矢量为基础进行设计,采用的查询算子是傅立叶描述子。实验结果表明,该傅立叶描述子计算的查询算子区分度高,查询效果好,是一种稳健高效的识别工具。

     

    Abstract: The recognization of the shapes of buildings,annotation and so on,is commonly based on grid data,whose research base point is pixel.The shortcomings of it lie in low precision and large calculations.A brand-new research thinking is put forward,which uses vector as research base point instead of grid(pixel) and designs recognization algorithm based upon vector.Polygonal inquiry based on shape template match is designed under such thinking.Inquiry operator in the method is Fourier descriptor.The experimental results show that inquiry operator using the new Fourier descriptor to calculate has high differentiation degrees and good query results,which is really a steady and highly effective recognization tool.

     

/

返回文章
返回