The recognition of structures in SAR imagery by the use of a spatial reasoning method

Abstract

ASBTARCT Structures with energy supply capacity are an important part of lifeline engineering. Post-disaster assessment is therefore vital to assess the economic losses caused by the collapse of such structures and is an integral part of the overall post-disaster assessment. The identification of structures in synthetic aperture radar (SAR) imagery is mostly influenced by buildings, exposed rocks and riverbeds, because of the similar scattering intensity effects. In this letter, a new spatial reasoning method is proposed based on spatial and morphological feature parameters to identify isolated structures with an ordered distribution in SAR imagery. The wind power generation base in the north-western part of the city of Yumen, Gansu province, China, is taken as the research area. Numerous wind power generation facilities, which are important power sources in north-western China, are found in this area. The recognition accuracy of the proposed method for structures in this area is 98.31%. Compared with the traditional method, the recognition rate for structures is improved by 35.59%. The ‘distance deviation risk’ evaluation index is also proposed to evaluate the deviation between the recognition results and the actual ground objects. Overall, the proposed method can more accurately identify the structures in SAR imagery.