Abstract：Road network extraction in SAR images is one of the key tasks of military and civilian technologies. To solve the issues of road extraction of HJ-1-C SAR images, a road extraction algorithm is proposed based on the integration of ratio and directional information. Due to the characteristic narrow dynamic range and low signal to noise ratio of HJ-1-C SAR images, a nonlinear quantization and an image filtering method based on a multi-scale autoregressive model are proposed here. A road extraction algorithm based on information fusion, which considers ratio and direction information, is also proposed. By processing Radon transformation, main road directions can be extracted. Cross interferences can be suppressed, and the road continuity can then be improved by the main direction alignment and secondary road extraction. The HJ-1-C SAR image acquired in Wuhan, China was used to evaluate the proposed method. The experimental results show good performance with correctness (80.5%) and quality (70.1%) when applied to a SAR image with complex content.
陆萍萍, 杜康宁, 禹卫东, 王 宇, 邓云凯. 基于特征融合的HJ-1-C SAR 图像道路特征提取算法[J]. 雷达学报, 2014, 3(3): 352-360.
Lu Ping-ping, Du Kang-ning, Yu Wei-dong, Wang Yu, Deng Yun-Kai. Feature Fusion Based Road Extraction for HJ-1-C SAR Image. JOURNAL OF RADARS, 2014, 3(3): 352-360.
