基于BP神经网络自动提取沟谷研究

 提出将提取沟谷的过程转化为根据地形因子综合判定地貌类型的思路。以陕西绥德黄土丘陵沟壑区域1∶1万地形图制作的分辨率为5 m的DEM为研究对象,运用BP神经网络分析6种地形因子与沟谷地形的相互关联关系,认为降水累积量是判定沟谷地形的最重要因子。在试验样区建立BP神经网络,利用4种地形因子自动提取沟谷,并在检验样区通过了检验。

Automatic extraction of channels and valleys based on BP neural network

The process of extracting channels and valleys is transfered to the process of determining the physiognomy type according to terrain factors.In Loess hill and gully area of Suide,Shannxi Province,DEM with spatial resolution at 5m according to the map scaled at 1∶10?000 is taken as the research object,and BP neural network is used to analyze the interrelationship between six kinds of terrain factors and the terrain of channels and valleys.The study indicated that precipitation accumulation is the most important terrain factor in determining the terrain of channels and valleys.The sample testing areas are established in which four kinds of terrain factors are made use of and BP neural network and employed to extract channels and valleys automatically.This method has passed the test in sample areas.