一种提高SAR目标识别率的有效方法

Efficient Approach of Improving SAR ATR Performance

在合成孔径雷达自动目标识别SARATR中,SAR像的预处理是提高识别率的关键技术之一。给出了一种简单有效的SAR图像预处理方法,该方法首先对SAR目标像进行对数变换后,再做傅立叶变换。经预处理后的SAR像用支持矢量机SVM分类器进行目标识别。实验结果表明:本方法不但有效地提高了目标识别率,而且保证了目标的平移不变性并具有良好的推广能力。

The preprocessing of SAR images is one of the key issues to improve SAR ATR performance.A simple and efficient approach of processing SAR images before recognition is presented in this paper.SVM classifier is used to implement target recognition after prepro-cessing.Experimental results showed a better performance of classification,generalization as well as a shift-invariance of target.