彩色图像中线状目标提取的透镜跟踪法

地形图扫描矢量化是矢量数据获取的主要途征,以往的方法主要是先将彩色扫描图二值化或者进行分层处理,再实施要素跟踪,这样容易造成要素断裂和数据的不完整。为了克服图像细化和分层的影响,在线状目标的半自动跟踪中,提出了一种基于后台区域细化,然后匹配跟踪的算法,即透镜跟踪法,实现了直接在24位扫描图像上的线划跟踪。该算法在多比例尺地形图上进行了大量数据采集实验,并成功运用到作者组织的数字化生产中,结果表明可以大大提高追踪速度和准确率,通过后期的处理,跟踪出的数据基本满足生产的需要,在实际生产中能发挥积极的作用。

Lens tracing method for linear objects skeleton from colored scanning map

Raster-to-vector conversion is the main method to get the vector data. In the past, we often switch the image to a binary image or distribute the plate of colored map before vector. Sometimes we collect the geographic data on the separated original data directly.But,it is inefficient,at the same time,it may lead to some consequences such as elements separation and data tattering. In order to remove the effects of map separating and thinning and extract the linear objects skeleton speedily and efficaciously from colored scanning map, this paper presents a detailed lens tracing method for liner objects based on its various appearances in image. An algorithm of part thinning and asynchronous operations is put forward. During the process of line tracing,a lens window is open and it moves along the line. Now, the contents in the window are only lines except the map background. We can put part-image into the memory of the computer and do some thinning. And then we can judge what is the middle and what is the intersection or the end of the line. According to the idea, we can extract the whole geometric information from the lines. Examination is done in many different liner objects conditions and in the actual map digitization, the system that takes the algorithm as the dominant factor is used by the author. It is proved that it is rapid and perfect to follow lines on a 24-bits scanning map in some actual data-extraction. It can be applied in many fields of data acquisition and data update of map surveying and GIS.