Intelligent Recognition Of Targets In Specific Areas Based On Satellite Remote Images

The rapid development of satellite image remote sensing technology has provided a large number of data sources for rapid and effective monitoring and extraction of natural resource information.With the development of new remote sensing technology,new technologies and challenges have been brought to the monitoring and statistical work of natural resources.Effectively grasping the distribution of various subdivisions of natural resources is of great significance to people’s production and life,and even has important guiding significance for the formulation of national policies.Based on high-resolution satellite remote sensing images of a specific area,this paper conducts parameter extraction and accompanying numerical calculation research for geographic information such as mountains,forests and fields.By analyzing the spectral features,texture features and shape features of the target,an effective extraction method for specific resource information such as mountains,forests and fields is proposed.The main research contents of this paper are as follows:1、Based on satellite remote sensing image data,the area of mountain forests in the region is divided and the area of mountain forests is calculated.On the basis of NDVI（Normalized Difference Vegetation Index）,a new vegetation index INDVI（Improved Normalized Difference Vegetation Index）is proposed for mountain forest extraction.Texture features are added to the forest extraction process,and the forest texture is extracted using a GLCM（Gray-level co-occurrence matrix）.The mountain forest extraction method using the combination of INDVI and GLCM outperforms methods such as threshold segmentation,NDVI,INDVI and GLCM.2、Based on satellite remote sensing images,the topography of farmland is extracted,and the fields are divided into paddy fields and dry fields for extraction research.For paddy fields,the NDWI（Normalized Difference Water Index）was improved to obtain the INDWI（Improved Normalized Difference Water Index）.And add the SWT（Stroke Width Transform）for text detection,and use the method of combining INDWI and SWT to extract paddy fields.For dry fields,three methods of threshold segmentation,K-means clustering and texture feature are used for discussion.By comparing the processing results of different regions,the K-means clustering method with higher accuracy is selected.3、A GUI visualization processing system is built,which can realize functions such as image cropping,mountain forest extraction and area calculation,field extraction and area calculation.