大型机场遥感图分割技术研究

发布时间：2019-01-05 10:51

【摘要】：图像分割是一项应用于图像领域的技术手段,也是当下研究人员研究图像处理方面一项关键性题目。至今,其方法仍在不断发展,比如基于边缘、区域、阈值以及聚类方法。然而针对大型机场遥感图的分割技术却研究甚少,单一的算法难以分割机场目标,所以多方法结合的分割研究,值得我们深入研究。本文针对大型机场遥感图像目标,进行分割研究,并将分割结果使用到实验室红外大场景建模中去,为后续项目的应用提供很好的前提准备。正如大家所熟知,不同的目标使用不同的分割法,会有不同分割效果,所以找到针对机场目标的分割方法尤为关键。考虑到机场遥感图中分割目标多,边缘参差不齐,外加航拍图像经常会受到天气等影响,呈现模糊等特点,传统方法很难得到好的分割结果。FCM算法是近年来应用比较广,针对边缘模糊、分割目标多,效果较好的方法,机场航拍图像非常符合其分割目标特征。因此,我们对FCM方法做一研究,对传统FCM方法进行改进,克服了传统FCM的初始聚类中心选定的随意性而导致的迭代次数多、效率差,以及传统的方法易受噪声影响的缺点,从而提高了算法的效率以及抗噪性,使得算法强大和实用性更强。多方法结合分割大型机场遥感图。首先,改进传统的FCM算法,对机场目标进行初分割,然后应用Canny算子,提取边缘,接着形态学半自动方式优化处理轮廓,最后将优化的轮廓与原始图匹配,提取出跑道目标;其次,进行实验验证本文算法的效率和普适性,并讨论未分割的建筑物群区域,提取出建筑物群;最后实现了多级分割机场目标。总结论文,阐述本文的优势和不足,以及后期还要努力的方向。实验结果表明:本文算法能得到良好的机场目标分割效果,易于实现和理解,现实意义很大。
[Abstract]:Image segmentation is a technical method applied in the field of image, and it is also a key topic in the field of image processing. Up to now, its methods, such as edge, region, threshold and clustering methods, are still being developed. However, there are few researches on the segmentation of large airport remote sensing images, and it is difficult for a single algorithm to segment the airport objects. Therefore, the multi-method combined segmentation research is worth our in-depth study. In this paper, the segmentation of large airport remote sensing images is studied, and the segmentation results are applied to the modeling of large infrared scene in laboratory, which provides a good premise for the application of subsequent projects. As we all know, different targets use different segmentation methods, there will be different segmentation effects, so it is very important to find a segmentation method for airport targets. Considering that there are many segmentation targets and uneven edges in airport remote sensing images, and the aerial images are often affected by the weather and appear fuzzy, it is difficult to obtain good segmentation results by traditional methods. FCM algorithm is widely used in recent years. Aiming at the method of edge blur, multi-target segmentation and better effect, the airport aerial image is very consistent with the segmentation target feature. Therefore, we study the FCM method and improve the traditional FCM method, which overcomes the disadvantages of the random selection of the initial clustering center of the traditional FCM, which results in a lot of iterations, low efficiency, and the traditional method is susceptible to noise. Thus, the efficiency and noise resistance of the algorithm are improved, which makes the algorithm more powerful and practical. Multi-method combined with segmentation of large airport remote sensing images. Firstly, the traditional FCM algorithm is improved to segment the airport target first, then the edge is extracted by using Canny operator, then the contour is optimized by morphological semi-automatic method. Finally, the optimized contour is matched with the original image, and the runway target is extracted. Secondly, experiments are carried out to verify the efficiency and universality of the proposed algorithm, and the undivided area of the building cluster is discussed to extract the building group. Finally, the multi-level segmentation of the airport target is realized. This paper summarizes the advantages and disadvantages of this paper, as well as the direction of the later efforts. The experimental results show that the proposed algorithm can achieve good segmentation effect and is easy to be realized and understood.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【参考文献】