Terrace Information Extraction Based On Texture Feature Of High Resolution Remote Sensing Image

As widely distributed on the slopes in hilly area and contour parallel step basic farmland,the construction of terraced fields considering the nature of the local landform features and the actual production and living needs,to fundamentally solve the contradiction of population and grain,with considerable economic benefits and social benefits,and terraces information access to important data basis is provided for regional layout of agricultural production.The techniques of analysis,classification and extraction for terrace information can be divided into four categories: visual interpretation,single pixel,object analysis and spectral texture.All kinds of methods have their own advantages and disadvantages.Due to the complexity and limitations,there are some difficulties and some inapplicability for different ground objects with difference in the practical application.In order to avoid the problems of poor classification accuracy,"same object with different spectrum" and " different objects with same spectrum " and noise interference,the texture information of remote sensing image was selected as the main research object and the classification basis to extract the terrace objects.Based on texture feature extraction method is not dependent on the images of the color and brightness,reflect the homogeneity phenomenon on the surface of the object in the image.It is helpful to further interpret the classification of ground objects.The feature extraction is necessity and superiority for terrace that texture feature prominent feature has certain,is also this article analyzed the important basis of object extraction terrace.This article is based on high resolution remote sensing image provided by Google Earth has typical texture terraces as the research object,from two aspects of space domain and frequency domain.According to analysis the spatial convolution filtering and Fast Fourier transform method,to sharpen the uniqueness of the terraces texture enhancement,restrain the influence of the redundant data and interference information and improve the accuracy and efficiency of object extraction.Based on the experiment of terrace extraction in a small area,a larger scale image was selected in the study area for terrace texture enhancement and terrace object extraction,The verification accuracy of the final extraction result reached more than 80%.Compared with the extraction of terrace based on spectral information,the overall extraction result was more ideal.From the perspective of enhancement effect,steep terrace have the best effect,followed by gentle terrace and flat terrace.However,the noise interference of steep terrace is more prominent during the extraction process,which makes the extraction result inferior to that of gentle terrace and flat terrace.To some extent,this study demonstrates the feasibility of extracting terrace information based on texture features of high-resolution images,and hopes to provide ideas for extracting terrace information in a wide range,improve the accuracy and efficiency of remote sensing image interpretation,solve the contradiction between highquality data and processing methods,and better the current big data environment,in order to better adapt to the current big data environment.