动态监测变化图斑边界跟踪

Tracing Edge of Change Plot for Dynamic Monitroing

在土地利用遥感动态监测中,自动发现的变化信息普遍基于彩色影像或灰度影像,而且变化信息图斑的灰度域值分布在一定范围内。为了适应快速提取变化信息,以及便于GIS软件管理监测数据的需要,在将包含变化信息的影像二值化的基础上,设计了一算法实现对变化信息图斑边界的自动矢量化提取。然后将得到的矢量图与不同方法得到的变化信息影像图进行叠加显示,通过多方法的对比,以对参与矢量化的变化信息影像图导致的图斑误判、边界误判进行纠正,提高数据处理精度。最后,通过与人工提取的变化信息边界进行对比及野外核实,证明了本方法是可行的,具有可靠的精度。

Change information, which automatically detected by land using dynamic monitoring, is commonly from color images or gray images, and its gray threshold is distributed in a certain range. In order to rapidly extract change information, and expediently manage monitoring data using GIS software, research should do to express change information with vector data automatically. Since image including change information contain complicated background, it is a difficult thing to extract directly edge of change information from the image, so it is necessary to segment change information from the image, which is a bi\|value process, then tracking vector edge for change information. The threshold segmentation arithmetic is used to bi\|value the image in this thesis, and an eight directions arithmetic is optimized to tracking edge for change information, and vector edge is judged to ensure it is inside or outside to confirm range of change information and to make sure topology relationship among edges. Last, displaying vector map and images containing change information and produced automatically in other methods together, so edit is done for vector map to enhance precision. Comparing to manual drawing vector data and verifying in field, our method is feasible, and data has reliable precision.