基于蓝噪声理论的遥感图像森林植被纹理研究

发布时间：2018-03-10 03:05

本文选题：遥感图像　切入点：蓝噪声　出处：《辽宁师范大学》2014年硕士论文　论文类型：学位论文

【摘要】：近年来，随着高分辨率遥感技术的迅速发展，高分辨率遥感影像可以针对面实体及其空间位置信息进行较好的观察。植被是指覆盖地表的所有植物群落的总称，与土壤、气候、地形及水状况等自然环境要素密切相关。植被是生态环境的最佳反应区域之一，也是生态系统的重要组成部分，它在生态系统中一个非常重要的角色。快速准确地获取植被信息，先后对于城市建设和绿地规划等方面都具有重要意义。但植被结构复杂，种类繁多，具有多尺度集聚以及空间分布的破碎性特征，因此一个好的描述特定尺度像素的森林植被特征的分类方法才可以获得较为合理的分类结果。为了实现这一目标，选择一个合适的尺度来识别遥感图像植被信息是整个算法的基础。在开始阶段，一种基于蓝噪声特征理论的遥感图像森林植被纹理的测量方法被用于对遥感图像森林植被进行尺度提取，与传统的方法相比减少了算法的时间复杂度。最优尺度的提取在整个研究中是非常重要的，因为最优的分割比例决定了森林植被的大小和信息提取的精度。要使分割后的多边形能将原遥感图像森林植被边界显示清楚，既不能太破碎，也不能太模糊。随着研究的深入，基于组合各类纹理特征的提取方法的纹理模型被不断提出，但大多数模型由于方法组合的复杂度高、时间代价过大，而不得不在纹理表达确切度上降低要求。缺少简洁且确切的结构元尺度来表达模型实现森林植被分割，是目前的主要问题。本文提出了一种基于蓝噪声理论支持下建立一种快速探测森林植被纹理尺度的方法。经过纹理尺度探测对区域图像选择并缩小持续探测蓝噪声特征，利用探测到的蓝噪声特征及该尺度确定结构元素的尺度，利用得到的结构元素进行形态学滤波处理实现遥感图像森林植被区域的分割。根据分析得到的森林植被的特定结构元素，然后对遥感图像进行击中或击不中变换处理，使用探测得到的尺度所设计的结构元对森林植被进行分割提取。进一步，在这些的基础上可以根据探测出的灌木尺度，，设定特定的结构元来进行灌木分割，这样在较为复杂的遥感图像中就可以清晰的提取森林植被与灌木。
[Abstract]:In recent years, with the rapid development of high-resolution remote sensing technology, high-resolution remote sensing images can be used to better observe the surface entities and their spatial location information. Vegetation is one of the best response areas of ecological environment, and it is also an important part of ecosystem. It plays a very important role in the ecosystem. Obtaining vegetation information quickly and accurately is of great significance to urban construction and green space planning, but the vegetation structure is complex and there are many kinds of vegetation. It has the characteristics of multi-scale agglomeration and fragmentation of spatial distribution. Therefore, a good classification method to describe the forest vegetation features with specific size pixels can obtain more reasonable classification results. In order to achieve this goal, selecting a suitable scale to identify the vegetation information of remote sensing images is the basis of the whole algorithm. A remote sensing image forest vegetation texture measurement method based on blue noise feature theory is used to extract forest vegetation from remote sensing image. Compared with the traditional method, the time complexity of the algorithm is reduced. The extraction of the optimal scale is very important in the whole research. Because the optimal segmentation ratio determines the size of forest vegetation and the accuracy of information extraction, the segmented polygon can display the original remote sensing image forest vegetation boundary clearly, neither too broken nor too fuzzy. With the development of the research, the texture model based on the combination of various texture features has been proposed, but most of the models because of the high complexity of the method combination, the time cost is too large. The lack of concise and exact structural element scale to achieve forest vegetation segmentation is the main problem. In this paper, a method based on blue noise theory is proposed to quickly detect the texture scale of forest vegetation. Through texture scale detection, the region image is selected and the blue noise features are reduced. Using the detected blue noise characteristics and the scale of the structural elements, Based on the analysis of the specific structural elements of the forest vegetation, the remote sensing image can be hit or hit. The forest vegetation is segmented and extracted by using the structure element designed by the detected scale. Furthermore, according to the detected shrub scale, a specific structure element can be set to segment the shrub, so that the forest vegetation and shrub can be extracted clearly in the more complex remote sensing image.
【学位授予单位】：辽宁师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【参考文献】