高空间分辨率遥感影像分割自适应尺度计算

发布时间：2018-04-15 03:11

本文选题：高空间分辨率遥感影像 + 分割自适应尺度　；参考：《北京建筑大学》2017年硕士论文

【摘要】：随着空间对地探测技术的发展,遥感卫星对地观测有了重大发展,特别是在高空间分辨率遥感影像(the High Spatial Resolution Remote Sensing Image,以下简称HSRRSI)方面,HSRRSI数据量急剧增长。但是对应的HSRRSI数据处理能力发展相对缓慢,严重制约着HSRRSI应用水平。面向对象的影像分析(Object-Based Image Analysis,以下简称OBIA)为HSRRSI的分割分类提供了一个全新的思路,影像多尺度分割是OBIA的关键,而分割过程中分割自适应尺度计算是影像多尺度分割的基础,直接影响着多尺度分割的精度和影像地物目标自动化提取的效率。现有的分割尺度参数计算方法对先验经验的依赖性较强,大多只适用于特定的遥感影像,本文在对现有尺度计算方法总结分析的基础上,提出了高空间分辨率遥感影像空域尺度计算以及空域尺度约束条件下的值域尺度计算,主要研究内容如下:(1)分析总结现有尺度计算法的应用范围及优缺点,指出了尺度参数计算未来发展的趋势是自适应多尺度的精细化尺度计算方法。(2)探讨了影像滤波对HSRRSI分割精度的影响,通过实验验证了合适的滤波方法可以在有效消除影像中噪声信息、增强影像中地物目标语义影像内部像素信息同质性的同时,很好的保留了影像边缘的细节信息,有效提高了HSRRSI的分割精度及效率。(3)由于传统的Canny边缘检测算法只能对单通道影像的边缘进行边缘检测,本文根据HSRRSI的特点,提出了改进的Canny矢量边缘检测算法,通过实验证明该算法可以有效提取HSRRSI的矢量边缘信息。(4)提出了基于矢量边缘的高空间分辨率遥感影像空域尺度计算方法及在空域尺度约束条件下的值域尺度计算方法,并利用均值漂移分割方法对值域尺度影像进行多尺度分割,验证了空域尺度计算与值域尺度计算的合理性。
[Abstract]:With the development of space-to-earth detection technology, there has been a great development in remote sensing satellite Earth observation, especially in the aspect of High Spatial Resolution Remote Sensing image (HSRRSI).However, the corresponding HSRRSI data processing ability is relatively slow, which seriously restricts the level of HSRRSI application.Object-Based Image Analysis (Object-Based Image Analysis) provides a new idea for the segmentation and classification of HSRRSI. Multi-scale image segmentation is the key of OBIA, and adaptive scale calculation is the basis of multi-scale segmentation.It directly affects the accuracy of multi-scale segmentation and the efficiency of automatic object extraction.The existing methods for calculating the parameters of segmentation scale are highly dependent on prior experience, and most of them are only suitable for specific remote sensing images. In this paper, based on the summary and analysis of the existing methods of scale calculation,The spatial scale calculation of high spatial resolution remote sensing image and the range scale calculation under the constraint of spatial scale are put forward. The main research contents are as follows: 1) the scope of application, advantages and disadvantages of the existing scale calculation method are analyzed and summarized.It is pointed out that the development trend of scale parameter calculation in the future is adaptive multi-scale finer scale calculation method. The influence of image filtering on HSRRSI segmentation accuracy is discussed.The experiments show that the proper filtering method can effectively eliminate the noise information in the image, enhance the homogeneity of the pixel information in the semantic image of the ground object, and at the same time preserve the details of the edge of the image.Because the traditional Canny edge detection algorithm can only detect the edge of single channel image, according to the characteristics of HSRRSI, an improved Canny vector edge detection algorithm is proposed in this paper.It is proved by experiments that the algorithm can effectively extract vector edge information from HSRRSI.) the spatial scale calculation method of high spatial resolution remote sensing image based on vector edge and the range scale calculation method under the condition of spatial scale constraint are proposed.The multi-scale segmentation of range scale image is carried out by means of mean shift segmentation method, which verifies the rationality of spatial scale calculation and range scale calculation.
【学位授予单位】：北京建筑大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P237

【参考文献】