Survey of Target Detection Algorithms in SAR Images

As a microwave remote sensing imaging radar, SAR has been widely used in military and civilian fields due to its advantages such as all-weather and all-day imaging. However, due to the exponential growth of image data, new SAR image target detection techniques are needed. Convolutional neural network provides a good idea for SAR target detection technology. For SAR images, the traditional target detection algorithm based on constant false alarm rate (CFAR) is firstly analyzed; then the target detection algorithm based on convolutional neural network is combed, including several algorithms based on candidate regions such as R-CNN and R-FCN, and several algorithms based on regression models such as YOLO and SSD. Finally, other target detection algorithms are introduced, the characteristics and the existing problems of various algorithms are analyzed. On this basis, the problems existing in SAR image target detection technology and further research are prospected.