The Study Of Object-oriented Coastline Extraction In Xiushan Island Based On GF-2 Remote Sensing Image

Coastline is an important boundary between land and sea,which presents different forms in the remote sensing image.The coastline is constantly changing due to the interaction of various geological factors,the change of sediment at the mouth of river,the influence of meteorological and oceanic conditions,the rise of sea level and the impact of human social and economic activities.For islands,coastline resource is an important natural resource,which is closely related to the production and life of island residents,it is the basis of existence for sea space resources.Therefore,how to realize the efficient and accurate extraction of island coastline will become the main research direction.In this paper,the study area is Xiushan Island,Daishan County,Zhoushan City Zhejiang Province,and GF-2 remote sensing image as the data source.The coastline of Xiushan Island is automatically extracted by object-oriented classification method,the obtained coastline is adjusted for tide level,and combined with the definition standard of artificial coastline in Zhejiang Province " Specification for coastline survey statistics",the parts of coastline is modified.Finally,the real coastline of Xiushan Island is obtained.The man contents of this paper are as follows:(1)The four types of coastlines in XiuShan Island,including bedrock,sandy,silt-muddy,artificial,are used to analyze the object characteristics and establish remote sensing interpretation signs.The image preprocessing of GF-2 remote sensing image includes radiation calibration,atmospheric correction,ortho rectification,geometric correction,image fusion,etc.and cutting the image according to the coastal type.(2)After preprocessing,the GF-2 remote sensing image is segmented in multi-scale.The segmented image is classified into water and non-water by using normalized difference water index(NDWI)and other parameters,combined with membership function rule in object-oriented classification.Finally,after visual interpretation to correct the error area in the classification process,the instantaneous waterline of the image is extracted.(3)By comparing with the coastline extracted by accurate manual visual interpretation,the accuracy of the coastline after tide level correction is evaluated,the root mean square error of the four types of coastlines is calculated.The root mean square error of artificial coastline is 0.89 m,the root mean square error of sandy coastline is 1.59 m,and the root mean square error of bedrock coastline is 3.11 m,the root mean square error of silt-muddy coastline is 20.10 m.The positions of the two coastlines are basically the same.(4)The automatically extracted waterline is vectorized,and some sections of the coastline are modified,finally the coastline data of Xiushan Island is obtained.Combining with the coastline type to analyze the coastline data,the total coastline length of Xiushan Island in 2019 is 40.32 km,the bedrock coastline length is 25.43 km,the sandy coastline length is 1.77 km,the artificial coastline length is 11.40 km,and the silt-muddy coastline length is 1.72 km,which is basically consistent with the coastline length extracted by visual interpretation.Through precision evaluation and data analysis,it is proved that the method used in this paper can be effectively applied to the extraction of island coastline,and provide technical support for the development and protection of island coastline.