Transmission Tower Extraction in High Resolution SAR Image Based on NSCT

Power transmission towers are the main infrastructure of power grid. It is important to monitor the safety state of transmission towers. Using synthetic aperture radar (SAR) image to monitor the transmission line is not limited by cloud, rain and snow and sunlight conditions. So it has been widely studied to detect transmission tower in a SAR image. However, most of the works consider the transmission tower as point target and few studies have focused on the extraction of fine structure of the tower. This paper proposes a SAR image target extraction algorithm based on NSCT domain feature extraction and fuzzy clustering. First, the characteristics of transmission tower in SAR image in NSCT domain is analyzed. Then, the proposed method extracts the statistical characteristics of each scale and each direction in NSCT domain as features. Fuzzy C means algorithm is applied to unsupervised features clustering and the final target extraction result is obtained. Finally, the validity of the proposed method is illuminated by the experiment using the TerraSAR-X images. Experiments show that the contour of the tower structure can be clearly identified in the extraction results.