Research On Automatic Identification And Extraction Of Terraces From China GF-1 Satellite Data

The terrace is one of effective measures for controlling soil and water loss on slope land,which has very significant function for water storage,soil conservation and increasing yield.In China terrace has thousands of year history,and at present terrace area is more than 20 million hm2.With the launch of comprehensive management engineering of transforming slope into terrace in China,area of terraces is expanding continuously.In order to understand the dynamic changes of terraces in a scientific way and ensuring the rational allocation of soil and water conservation measure,it is important to get timely and accurate inventory of spatial distribution of existing terraces.The remote sensing technology has the advantages in quickness,high accuracy and good freshness.This study uses edge detection to extract edge lines as the research object and proposes an edge characteristics statistics method,as well as Fourier transformation algorithm has been used,to carry out automatic identification and extraction study in terraces on high resolution satellite data(China GF-1).The results have shown that edge characteristics statistics method is based on terraced texture extraction of the edge lines,which is suitable for the identification and extraction of typical terraces.In this study,the study area has been selected in Hengshan County,Yulin of Shanxi Province.Field validation of terraces automatic identification,as well as extraction algorithm has been carried out here.The result has shown that the accuracy of terraces automatic identification and extraction can reach over 80%and the consumption of time is 50 minutes.