论文部分内容阅读
目标探测是遥感影像信息提取中的重要内容,然而,随着目标像元数目增多和相似地物的干扰,目标探测的虚警率会明显上升。将线性约束最小方差方法(LCMV)与局部对比方法(LCM)相结合,构建了一种新的多光谱遥感图像中目标探测方法(LCLCM):首先利用样本相关矩阵对目标进行半解混,然后利用图像的空间性增强目标信息、抑制背景信息,最后进行图像归一化和图像分割。以Landsat 8多光谱图像中船只提取为例进行方法验证,LCLCM的虚警率为1.07%,优于LCMV和LCM的虚警率12.39%和11.26%,表明该方法能够进行有效稳健的目标探测。
Target detection is an important part of remote sensing image information extraction. However, with the increase of the number of target pixels and the interference of similar objects, the false alarm rate of target detection will increase obviously. In this paper, we propose a new target detection method (LCLCM) for multispectral remote sensing images by combining linear constrained least square method (LCMV) and local contrast method (LCM). Firstly, the target is semi-unmixed with sample correlation matrix and then The spatial information of the image is used to enhance the target information and suppress the background information. Finally, image normalization and image segmentation are carried out. The method was validated by an example of vessel extraction in Landsat 8 multispectral imagery. The false alarm rate of LCLCM was 1.07%, which was 12.39% and 11.26% better than that of LCMV and LCM respectively. It shows that this method can effectively and robustly detect targets.