CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleResearch on Detection Methods of Ship Targets in Infrared Images with Sea-sky Background
Thesis Advisor薛文芳
Degree Grantor中国科学院大学
Place of Conferral中国科学院自动化研究所
Degree Discipline计算机应用技术
Keyword舰船目标检测 海天线检测 显著性分析 红外图像 Ship Target Detection Sea-sky Line Detection Saliency Analysis Infrared Images
Abstract海天背景红外图像舰船目标检测能在海战中提前发现敌军舰船,为火控系统提供目标方位指示。除此之外,海天背景红外图像舰船目标检测在捕鱼船监控,海上搜救等民事领域里也有着广泛应用。现阶段的舰船目标检测方法在复杂环境的适应性和检测准确率方面有待提高。基于实际应用场景,本文主要研究舰载红外图像的中小型舰船目标检测方法。在此限制条件下,当图像中存在海天线时,舰船目标通常位于海天线附近。因此,在海天线附近区域检测舰船目标,能减少背景干扰,提高检测准确率并降低计算量。本文将海天背景红外图像舰船目标检测分为海天线检测和舰船目标检测两部分。主要工作如下: (1)设计了一种基于链码的海天线检测方法。该方法首先对原始图像下采样,降低图像的分辨率,减少后续步骤的计算量;然后通过垂直方向的梯度算子生成边缘图;接着采用一种新的基于海天线特性的链码跟踪方法跟踪链码并在链码中检测直线;最后判别检测得到的直线中是否存在海天线。经过实验比较发现,本文设计的海天线检测方法对复杂多变的海天背景具有良好的适应性,检测准确率高。 (2)实现了一种基于显著性分析的舰船目标检测方法。该方法融合了基于全局上下文和局部上下文的显著性检测结果。当海天背景红外图像中未检测出海天线时,该方法还融合了一种新的基于背景先验知识的显著性检测结果。实验结果显示,相比一些经典的基于显著性分析的目标检测方法,在海天背景红外图像舰船目标检测这一应用领域中,本文实现的方法具有更好的检测效果。
Other AbstractThe detection of ship targets in infrared images with sea-sky background can find enemy ships in advance and provide position indication for fire control system. Furthermore, it is also very useful in many civil fields such as fishing vessel surveillance and maritime search and rescue. Nowadays, there is still a large gap between existing ship target detecting methods in infrared images with sea-sky background and the utility. Many aspects need to be improved, such as environment adaptability and detection accuracy. According to actual application scenarios, this paper mainly researchs on the detection of medium and small ship targets in shipboard infrared images. Under this limited ccondition, it can be found that when the sea-sky line appears in the image, ship targets often appear in the area near the sea-sky line. Therefore, we only need to detect ship targets in the neighborhood of the sea-sky line. Then, the calculation amount of detection algorithm could be largely reduced for the diminution of the background interference, with an improvement of detection accuracy. In this paper, we divide ship targets detection in infrared images with sea-sky background into two parts: sea-sky line detection and ship targets detection. The main contributions are summarized as follows. (1) A sea-sky line detecting method based on chain code is designed. In this method, we firstly downsample the original image in order to reduce the image resolution and the calculation amount of subsequent steps. Secondly, edge image is generated according to vertical gradient. Then, a new chain code following method based on the characteristics of sea-sky line is used and a line detecting method based on chain code is applied to detect lines in the edge image. Finally, whether the sea-sky line exists in those detected lines will be judged. Experimental comparison shows that, our method has good adaptability to complex sea environment and high detection accuracy. (2) A ship target detecting method based on saliency analysis is realized. This method combines results of global-context saliency detection and local-context saliency detection. Futhermore, when the sea-sky line is not detected, this method also combines results of a new saliency detection method based on background prior knowledge. Experimental results show that in the application of ship target detection in infrared images with sea-sky background, our method has better detection results compared with some classi...
Other Identifier201228014629101
Document Type学位论文
Recommended Citation
GB/T 7714
吴芳. 海天背景红外图像舰船目标检测方法研究[D]. 中国科学院自动化研究所. 中国科学院大学,2015.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20122801462910(2483KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[吴芳]'s Articles
Baidu academic
Similar articles in Baidu academic
[吴芳]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[吴芳]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.