A detection method for SAR targets based on combining multiple features is proposed. The targets of interest are detected according to the physical properties, which reflect the true characteristics including scattering intensity, size and differences from the clutter. By analyzing these characteristics, the size and boundary changes are determined as effective features. The image background, natural clutter, man-made clutter are eliminated in sequence using the developed detection algorithm, which contains two layers, namely, the initial target detection layer and the potential target identification layer. Effective features ensure that a smaller number of features are used to meet the precision of the target detection, and the discrimination detection method ensure that the probability of false alarm is reduced gradually with the increased complexity of the feature extraction. Comparison with traditional target detectors, such as CFAR, PCA, etc. is performed in detail. Experimental results show the superiorities of the proposal in both accuracy and efficiency.
