SAR TARGET RECOGNITION COMBINING CLASS INFORMATION AND CLASSIFICATION HYPERPLANE

Automatic target recognition (ATR) in synthetic aperture radar (SAR) images plays an important role. An important step in SAR target recognition is to transform the raw data to a feature space with the property of within-class compactness and between-class separation. However, due to the problem of target aspect angle sensitivity in SAR images, the difference of some targets belonging to the same class in different poses is greater than the difference of targets belonging to different class in similar poses, which often causes misjudgement. To solve this problem, we first use the twin support vector machine (TWSVM) to obtain the classification hyperplane of each class. Then, based on the obtained hyperplane, the projection matrix is constructed by fusing the prior class information of the target. In this way, the sample can be close to the classification hyperplane of the same class and distant from the classification hyperplane of a different class. This algorithm can improve TWSVM's ability to classify SAR images. Experiments are conducted on the moving and stationary target acquisition and recognition (MSTAR) database. The results verify the effectiveness of the proposed algorithm.