目录
摘要II
关键词II
AbstractIII
引言
引言1
1绪论1
1.1研究背景和意义1
1.2国内外研究现状1
1.3研究内容3
1.4章节安排4
2遥感图像数据集制作4
2.1 DOTA数据集 4
2.2人工标注数据 5
2.3数据集转为VOC2007格式 6
3 三种目标检测方法研究9
3.1 FPN目标检测算法9
3.1.1算法原理 9
3.1.2训练和测试 11
3.2 YOLOv3目标检测算法15
3.2.1算法原理 15
3.2.2训练和测试 18
3.3 RefineDet目标检测算法22
3.3.1算法原理 22
3.3.2训练和测试 24
3.4三种目标检测方法的比较29
4 Web演示系统搭建30
4.1系统设计30
4.2系统演示32
5总结与展望36
致谢38
参考文献39
基于遥感图像的目标检测方法研究
摘 要
随着遥感技术的飞速发展，高分辨率的遥感图像的数量不断增加，使得对遥感图像进行的目标检测成为可能。目标检测的任务就是在给定的图像中精确定位目标并判断目标所属类别。遥感图像中的目标尺度相差大，这要求应用在遥感图像上的目标检测算法不仅要能检测出较大的目标，还要对较小的目标有良好的检测效果。然而，现有的目标检测算法大多对大目标有较好的检测效果，但对小目标的检测效果不是很理想。故研究基于遥感图像的目标检测算法具有重要意义。首先在DOTA数据集的基础上结合人工标注的遥感图像制成了遥感图像数据集。然后选择了三种目标检测算法，分别为Two stage类型的FPN算法、One stage类型的YOLOv3算法和混合类型的RefineDet算法，将三种目标检测算法在遥感图像数据集上进行训练和测试。经过实验验证对比，RefineDet算法的平均精度均值mAP最 *51今日免费论文网|www.51jrft.com +Q: #351916072# 
高，对遥感图像中的大目标和小目标的检测都有着较好的效果。之后总结了各算法的优缺点并提出了可能的改进方法。最后实现三个算法目标检测效果的可视化展示，将训练好的目标检测模型部署在Web网站上进行演示。
RESEARCH ON TARGET DETECTION METHOD BASED ON REMOTE SENSING IMAGE
ABSTRACT
With the rapid development of remote sensing technology, the number of highresolution remote sensing images is increasing, which makes the target detection of remote sensing images possible. The task of target detection is to accurately locate the target in a given image and determine the category of the target.The difference of target scale in remote sensing image is large, which requires that the target detection algorithm applied in remote sensing image not only can detect larger target, but also has good detection effect on smaller target. However, most of the existing target detection algorithms have good detection effect for large targets, but the detection effect for small targets is not very ideal. Therefore, it is of great significance to study the target detection algorithm based on remote sensing image.Firstly, the remote sensing image data set is made based on the DOTA data set and the manually labeled Remote sensing image. Then three target detection algorithms are selected, which are two stage type FPN algorithm, one stage type yolov3 algorithm and hybrid type refinedet algorithm. Three target detection algorithms are trained and tested on remote sensing image data set. Through experimental verification and comparison, the average precision of RefineDet algorithm is the highest, which has a good effect on the detection of large and small targets in remote sensing image. Then, the advantages and disadvantages of each algorithm are summarized and the possible improvement methods are proposed. Finally, the visual display of the three algorithms is realized, and the trained target detection model is deployed on the web site for demonstration.

原文链接：http://www.jxszl.com/jsj/wljs/606911.html