基于频谱分布量化分析的星载人工目标快速检测方法
发布时间：2018-08-07 21:39
【摘要】：地物特征不单靠空域中像元灰度值的变化程度来表现,其频谱能够表征不同地物的特征,并可通过该频谱特征进行遥感图像的分析和目标的搜索、检测。提出一种通过频谱分析快速发现自然背景中人工目标及其所在区域的新方法,首先对图像的频谱分布特征进行了分析并提出了一种频谱分布量化的算法,定义频谱分布量化指数;根据计算出的区域频谱分布量化指数,借鉴视觉显著性的思想计算出各区域在整幅图像中的显著值,从而实现人工目标的检测。文中对该方法在海背景中舰船目标的检测和沙漠、戈壁中人工建筑的检测进行了实验,从结果可以看出新算法能够有效地实现自然背景潜在人工目标的快速搜索和发现,并确定出其所在区域,具有一定的灵活性,能够适应星上多变的检测场景。
[Abstract]:The feature of ground object is not only represented by the change degree of pixel gray value in the spatial domain, but also can be represented by the spectrum of different features of the ground object, and it can be used to analyze the remote sensing image and search for the target through the spectrum feature. In this paper, a new method to quickly find artificial objects and their regions in natural background by spectrum analysis is proposed. Firstly, the spectrum distribution characteristics of images are analyzed and an algorithm for quantization of spectrum distribution is proposed. The quantization index of spectrum distribution is defined and the significant value of each region in the whole image is calculated according to the calculated quantization index of regional spectrum distribution using the idea of visual saliency so as to realize the detection of artificial target. In this paper, the detection of ship targets in sea background and desert, and the detection of artificial buildings in Gobi are tested. The results show that the new algorithm can effectively realize the fast search and discovery of potential artificial targets in natural background. The region is determined, which is flexible and can adapt to the changeable detection scene on the satellite.
【作者单位】： 北京空间飞行器总体设计部;北京航空航天大学;
【基金】：国家自然科学基金资助项目(91438102)
【分类号】：TP751
本文编号：2171428
[Abstract]:The feature of ground object is not only represented by the change degree of pixel gray value in the spatial domain, but also can be represented by the spectrum of different features of the ground object, and it can be used to analyze the remote sensing image and search for the target through the spectrum feature. In this paper, a new method to quickly find artificial objects and their regions in natural background by spectrum analysis is proposed. Firstly, the spectrum distribution characteristics of images are analyzed and an algorithm for quantization of spectrum distribution is proposed. The quantization index of spectrum distribution is defined and the significant value of each region in the whole image is calculated according to the calculated quantization index of regional spectrum distribution using the idea of visual saliency so as to realize the detection of artificial target. In this paper, the detection of ship targets in sea background and desert, and the detection of artificial buildings in Gobi are tested. The results show that the new algorithm can effectively realize the fast search and discovery of potential artificial targets in natural background. The region is determined, which is flexible and can adapt to the changeable detection scene on the satellite.
【作者单位】： 北京空间飞行器总体设计部;北京航空航天大学;
【基金】：国家自然科学基金资助项目(91438102)
【分类号】：TP751
【相似文献】
相关期刊论文 前1条
1 闵德芬;;谈谈频谱分析[J];物理教学;1982年02期
，本文编号：2171428
本文链接：https://www.wllwen.com/guanlilunwen/gongchengguanli/2171428.html
