论文部分内容阅读
以黄土沟壑区甘肃天水桥子沟流域为例,研究乡村道路在高分辨率影像下的最优尺度分割和评价,基于Definiens平台面向对象的分割算法,将对象内部像素及灰度共生矩阵(GLCM)纹理的标准差的和作为衡量对象内同质性的标准,空间自相关系数为对象间的异质性度量标准,建立了一个最优分割尺度目标函数,对World View-2全色影像的2 876×3 778个像素,在1~100分割尺度范围内进行试验和循环计算。研究结果表明:三类道路对应的最优分割尺度分别为20、28、43,其中在分割尺度20、28、43下,对应的三类、二类和一类道路的差异性距离相比其他2类道路的差异性距离为最小,分别为0.031 7、0.034 4、0.016 8,此结果与最优目标函数计算结果吻合;利用目标函数法得到的分割尺度,在实际对乡村道路的最优尺度分割中是可行的,同时差异距离指数是一个较好的评价最优分割状态的指标。
Taking Qiaobagou watershed of Tianshui, Gansu Province as an example, the optimal scale segmentation and evaluation of rural roads under high-resolution images is studied. Based on the object-oriented segmentation algorithm of Definiens platform, the internal pixels and gray level co-occurrence matrix (GLCM) ) Texture standard deviation as a measure of homogeneity within the object, the spatial autocorrelation coefficient for the heterogeneity of objects between the metrics, the establishment of an optimal segmentation objective function of the scale, the World View-2 panchromatic image 2 876 × 3 778 pixels, in the range of 1 ~ 100 segmentation test and cycle calculations. The results show that the optimal segmentation scale corresponding to the three types of roads are 20, 28 and 43, respectively, and the corresponding distances between the three types, two types and one type of roads under the segmentation scale of 20, 28 and 43 are lower than those of the other The difference between the two types of roads is the smallest, which are 0.031 7,0.034 4,0.016 8 respectively. The result is in good agreement with the optimal objective function. Based on the segmentation scale obtained from the objective function method, Segmentation is feasible, while the distance difference index is a better indicator to evaluate the optimal segmentation status.