基于遥感影像的地表空间结构变异函数探测分析

发布时间：2018-05-25 00:07

本文选题：遥感影像 + 景观格局　；参考：《中国科学院大学(中国科学院东北地理与农业生态研究所)》2017年硕士论文

【摘要】：随着多光谱传感器的广泛运用,利用地物光谱响应特征提取地表信息的技术日益成熟。但由于地表状况的复杂性和光谱响应的局限性,传统的光谱分析法在指示平均尺寸、空间分布等格局信息方面存在缺陷。已有研究发现变异函数参数与地表空间结构参数存在一定的对应关系,但由于当前变异参数和场景参数关系的研究大都基于真实影像,不具备完整性和系列性。对变异函数探测性能缺乏全面分析,使得目前变异函数探测遥感影像树冠尺寸的效果仍然存在疑问。本文基于一系列不同尺寸分布和空间分布的一维和二维模拟影像,计算其变异函数值并提取变程。通过分析变异函数变程与模拟地物尺寸之间的关系,探究影响变程探测地物尺寸效果的因素,以及不同空间结构条件下变程探测地物大小准确性。最后结合法国南部地中海沿岸和西班牙中部的真实影像对模拟影像得出的研究结论进行验证。研究结果表明:⑴当森林树冠直径小于树冠之间平均距离一半时,且树冠尺寸变动方差较小时,利用变异函数变程参量可以有效探测出遥感影像中森林树冠尺寸;⑵当森林树冠直径大于树冠之间平均距离一半时,变异函数变程参量基本不随树冠尺寸增大而变化;⑶当森林树冠尺寸方差较大且树冠直径小于树冠之间平均距离一半时,变异函数曲线的拐点探测出的树冠尺寸要小于其真实值,而峰值变程探测出的树冠尺寸要大于其真实值,并且这种偏差随着树冠尺寸方差的增大而增加。
[Abstract]:With the wide application of multispectral sensors, the technology of extracting surface information by using spectral response features of ground objects is becoming more and more mature. However, due to the complexity of the surface condition and the limitation of spectral response, the traditional spectral analysis method has defects in indicating the average size, spatial distribution and other pattern information. It has been found that there is a certain correspondence between the parameters of variation function and the parameters of surface spatial structure. However, most of the researches on the relationship between variation parameters and scene parameters are based on real images, so they do not have integrity and series. Due to the lack of comprehensive analysis on the detection performance of variogram, the effect of variogram in detecting the crown size of remote sensing images is still in doubt. Based on a series of one-dimensional and two-dimensional simulated images with different size distribution and spatial distribution, the variogram value is calculated and the variable range is extracted. By analyzing the relationship between variogram variation and simulated object size, the factors that affect the effect of variable-range detection and the accuracy of variable-range detection under different spatial structure are explored. Finally, real images of the Mediterranean coast of southern France and central Spain are used to verify the results of the simulated images. The results show that when the diameter of forest crown is less than half of the average distance between tree crown and the variance of crown size variation is small, the variation parameter of variation function can effectively detect the forest crown size in remote sensing image. 2When the diameter of forest tree crown is larger than half of the average distance between tree crown, the variable parameter of variation function basically does not change with the increase of crown size. 3 when the diameter of forest tree crown is less than half of the average distance between tree crown, the variance of tree crown size is larger than that of tree crown. The crown size detected at the inflection point of the variogram curve is smaller than its true value, while the crown size detected by the peak variation is larger than its true value, and the deviation increases with the increase of the crown size variance.
【学位授予单位】：中国科学院大学(中国科学院东北地理与农业生态研究所)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P237

【参考文献】