约束最大相关系数的高光谱影像目标探测研究

发布时间：2018-05-23 11:58

本文选题：目标探测 + 相关系数　；参考：《激光与红外》2016年01期

【摘要】：高光谱遥感影像具有高的空间分辨率和连续的光谱信息,在目标探测领域具有独特的优势。基于高光谱影像的目标探测技术是遥感理论与应用的重要领域之一。本文从统计学中的相关系数的概念出发,提出了基于约束最大相关系数的高光谱影像目标探测算法。利用高光谱影像的线性混合模型,在真实图像中添加目标光谱,获得不同含量的亚像素目标及大目标,利用实验室高光谱成像仪对大目标进行推扫成像获取真实大目标高光谱影像。对仿真图像与真实图像进行约束能量最小化算子和约束最大相关系数算子进行对比,实验结果表明,基于约束最大相关系数的高光谱影像目标探测算法在探测大目标中具有更稳健的探测性能。
[Abstract]:Hyperspectral remote sensing images with high spatial resolution and continuous spectral information have unique advantages in the field of target detection. Target detection based on hyperspectral image is one of the important fields of remote sensing theory and application. Based on the concept of correlation coefficient in statistics, an algorithm for target detection in hyperspectral images based on constrained maximum correlation coefficients is proposed in this paper. Using the linear mixed model of hyperspectral image, the target spectrum is added to the real image, and the sub-pixel target and large target with different content are obtained. The hyperspectral images of real large targets are obtained by push-scan imaging with laboratory hyperspectral imagers. The constrained energy minimization operator and the constrained maximum correlation coefficient operator are compared between the simulation image and the real image. The experimental results show that, The hyperspectral image target detection algorithm based on constrained maximum correlation coefficient has more robust detection performance in detecting large targets.
【作者单位】：西安石油大学计算机学院;中国科学院西安光学精密机械研究所光谱成像重点实验室;中国人民武装警察部队工程大学;天津工业大学电子信息工程学院;
【基金】：国家自然科学基金项目(No.41301382;No.41301480;No.61401439) 教育部人文社会科学研究青年基金项目(No.14YJCZH172) 江苏省自然基金项目(No.20142BAB207007) 西安石油大学创新基金项目(No.YS29031606)资助
【分类号】：TP751

【相似文献】