海量灰度图像8连通域标记算法的设计与应用

发布时间：2018-04-12 10:10

本文选题：遥感影像数据 + 灰度图像　；参考：《河南大学》2014年硕士论文

【摘要】：在对遥感产品——即影像数据，如景观格局指数产品、城市环境遥感应用产品、国家级自然保护区动态监测与评价产品、全国生态质量遥感监测与评价产品、国家生态建设区域效果综合评价产品、全球环境变化监测与评价产品等——进行生产处理时，需先对遥感影像数据进行分割目标识别来实现信息提取，而连通域标记是分析分割结果以及进行后续目标识别的必要步骤。那么，如何对遥感影像数据进行连通域标记才能满足连通域间边界光滑、图像显示效果好的需求，是一个重要的问题。为此，对遥感影像数据的连通域标记从4连通增加到8连通，似乎可以达到连通域间边界光滑、图像显示效果好的目的。然而，遥感影像数据是一种灰度图像，数据量级较大，，如此计算量会大幅增加，算法效率将会严重下降。因此，寻求一种针对海量灰度图像的高效标记算法成为其中的关键问题。为了解决上述问题，本文提出一种海量灰度图像8连通域标记算法，并研究如何将该算法应用到遥感信息产品生产分系统原型系统的实际开发中。以下是本文主要研究内容和贡献。（1）提出一种海量灰度图像8连通域标记算法。针对现有连通域标记算法不能适用于遥感影像数据8连通域的快速标记问题，在现有的连通域标记算法基础上，对传统的8连通域标记算法进行八向变四向优化，采用列表机制解决连通域标记冲突问题，并引入了筛选机制和分块处理机制，实现了对海量灰度图像8连通域标记的快速处理，同时也得到了灰度图像连通域的正确划分及精确的连通域数目。实验结果证明，本文提出的海量灰度图像8连通域标记算法比现有算法的效率高、标记冲突次数少，更能适应海量灰度图像处理的要求。（2）将本文提出的海量灰度图像8连通域标记算法应用到国家重大专项课题之一“遥感信息产品生产分系统原型系统”实际开发中。结合实际的系统设计和开发过程，将该算法有效地应用到项目开发中，成功地解决了遥感影像数据连通域标记的问题，使得系统较好地实现对遥感影像数据进行分割目标识别来实现信息提取，正常进行景观格局指数产品、城市环境遥感应用产品、国家级自然保护区动态监测与评价产品、全国生态质量遥感监测与评价产品、国家生态建设区域效果综合评价产品、全球环境变化监测与评价产品等相关遥感产品的加工生产。
[Abstract]:In terms of remote sensing products-that is, image data, such as landscape pattern index products, urban environment remote sensing applications products, national nature reserve dynamic monitoring and evaluation products, national ecological quality remote sensing monitoring and evaluation products,When the products of comprehensive evaluation of regional effects of national ecological construction, global environmental change monitoring and evaluation products, etc. are processed, it is necessary to segment the remote sensing image data to realize information extraction.The connected domain labeling is a necessary step to analyze the segmentation results and to identify the subsequent targets.So, how to mark the connected domain of remote sensing image data can meet the needs of smooth boundary between the connected regions and good image display effect, which is an important problem.Therefore, the connected domain label of remote sensing image data increases from 4 connectivity to 8 connectivity, which seems to achieve the goal of smooth boundary between connected regions and good image display effect.However, remote sensing image data is a grayscale image, which has a large amount of data, so the computational complexity will be greatly increased, and the efficiency of the algorithm will be seriously reduced.Therefore, it is a key problem to find an efficient marking algorithm for massive gray images.In order to solve the above problems, this paper proposes an 8-connected domain labeling algorithm for massive gray images, and studies how to apply the algorithm to the development of prototype system of remote sensing information production subsystem.The following are the main contents and contributions of this paper.1) an 8-connected domain labeling algorithm for massive gray images is proposed.The existing connected domain labeling algorithms can not be applied to the fast labeling problem of 8-connected domain of remote sensing image data. Based on the existing connected domain labeling algorithms, the traditional 8-connected domain labeling algorithm is optimized in eight directions and four directions.The list mechanism is used to solve the conflict problem of the connected domain label, and the filtering mechanism and the block processing mechanism are introduced to realize the fast processing of the 8-connected domain label of the massive gray image.At the same time, the correct division and accurate number of connected domains of gray image are obtained.Experimental results show that the proposed 8-connected domain labeling algorithm for massive gray images is more efficient than the existing algorithms, and the number of labeling conflicts is less, so it can better meet the requirements of massive gray image processing.2) the 8 connected domain labeling algorithm of massive gray images is applied to the actual development of the prototype system of remote sensing information production subsystem, one of the major national projects.Combined with the actual system design and development process, the algorithm is effectively applied to the project development, and the problem of remote sensing image data connected domain marking is solved successfully.It makes the system realize segmentation target recognition of remote sensing image data to realize information extraction, normal landscape pattern index product, urban environment remote sensing application product, dynamic monitoring and evaluation product of national nature reserve.Processing and production of remote sensing monitoring and evaluation products for national ecological quality, comprehensive evaluation products for regional effects of national ecological construction, global environmental change monitoring and evaluation products and other related remote sensing products.
【学位授予单位】：河南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【相似文献】