Extraction Of Cultivated Land Information In Hilly Area Of East Sichuan Based On GF-2 Satellite Image

Cultivated land resources is the indispensable material basis for human survival and development.It is the root of grain self-sufficiency in the region.It is the reliable guarantee for national food security.Our country must have 1.8 billion mu of cultivated land.However,with the continuous development of urbanization,and demand for land resources is growing,cultivated land resources in this process is gradually occupied in this process.The sharp decrease of cultivated land area is the most significant feature of land use / cover change(LUCC)during this period..In terms of time and space,increase of the construction land means decrease of the cultivated land resources.Therefore,how to obtain farmland information quickly and accurately is great significance for the dynamic monitoring of cultivated land,the protection of cultivated land and the utilization of land resources,and to coordinate the relationship between human and land.It also can provide the basis for the formulation of relevant policies for the state.Because of same object with different spectra,different objects with same spectrum which reduced the accuracy of extraction of information.And because the cultivated land intertwined with the adjacent construction land,woodland and other objects,it formed the mixture which is difficult to distinguish.Difference of soil properties,crop and farming method and difference between the spectral characteristics of the class,which make the effect of information extraction based on medium and low resolution satellite image is unsatisfactory.With the rapid development of remote sensing and satellite technology,especially the development of high spatial resolution satellites,it is becoming possible to classify with high-precision.At present,using the high-resolution remote sensing image to extract information of cultivated land has becoming a research hot spot.Which provides a new idea for the rapid and effective extraction of large-area,high-precision remote sensing information,especially cultivated land information.In this paper,the farmland information is extracted from Longshi Town,Longchang City,which is located in the hilly area of eastern of Sichuan province.The town is the basic farmland protection demonstration town in Longchang City,and area of the cultivated land is large and representative.In this paper,the minimum distance method,maximum likelihood method,K mean method,ISODATA methodwhich are based on pixel and membership function classification method,CART decision tree method and support vector machine which are based on object-oriented are used to extract farmland information of the GF-2 satellite image in sample region.The main achievement is as follows(1)By means of Gram-Schmidt,PCA transform and NNDiffuse,fusing the GF-2images of the study region.And getting the best fusion method of the region imagines which is NNDiffuse by comparing the feature of brightness,contrast and texture.(2)Based on the spectral characteristics of the fusion images.,conducting the unsupervised experiments and supervised classification experiments.In the eCognition software platform,carrying out the multi-scale segmentation and determining the segmentation scale and related parameters suitable for imagine of the study area.The classification system is established.The feature is selected by feature optimization.Extracting the cultivated land information combining with the digital elevation model(DEM)of the study region.The results show that the classification effect based on the optimized feature is better than that only based on the spectral feature.(3)Combining the field verification,evaluating the accuracy of different methods to extract the information of cultivated land,the results show that the object-oriented classification method can extract the cultivated land information in hilly area more quickly and accurately.(4)Analyzing the relationship between the distribution of cultivated land and the terrain slope in the study area,and giving some relevant suggestion on the protection of cultivated land and the basic farmland.