论文部分内容阅读
维数约简是目标识别的一个重要预处理步骤。由于飞机目标图像对各种空间变换(包括平移、尺度、旋转等变换)和观察角度、位置以及光照等因素都比较敏感,使得很多线性维数约简算法不能有效地用于飞机目标识别。局部线性嵌入(LLE)是一种有效的非线性维数约简方法,提出了一种基于LLE的监督LLE算法,并应用于多种条件下的飞机目标识别中。实验结果表明,该方法是有效可行的。
Dimension reduction is an important preprocessing step in target recognition. Because of the sensitivity of aircraft target images to various spatial transformations (including translation, scaling, rotation, etc.) and observation angles, positions and illumination, many linear dimensionality reduction algorithms can not be effectively used for aircraft target recognition. Local linear embedding (LLE) is an effective nonlinear dimension reduction method. A LLE-based supervised LLE algorithm is proposed and applied to aircraft target recognition under various conditions. Experimental results show that this method is effective and feasible.