遥感数据的分形测量

用分形几何学方法对遥感数据进行了分析，从而定量解释遥感影像。共编制了３种分数维测量方法，它们是分线方法（或等值线方法），方差方法和三角棱柱方法。用分线方法计算了腾冲地区陆地卫星ＴＭ数据的分数维。图像数据的等值线长度分别按行，按列和按行－列方法处理记取以便能发现图像的模式变化。当图像有明显的走向模式时，按行方法所测量的Ｄ值与按列方法测得的Ｄ值是不同的。按行－列方法测量的Ｄ值几乎不受定向模式的影响。当Ｒ（相关系数）＞０．９时，按３种方式记取等值线所得到的Ｄ值的变化趋势是一致的，都能给出图像数据中的空间相互关系。通过分维测量，不但能描述图像数据的整体特征而且能描述图像数据中的每个值的特征，揭示遥感数据表观的无序性和内在的规律性，更重要的是可把这些Ｄ值与地面物体之间建立联系来探讨各种自然现象中的有用信息。

Fractal Measurement of Remote Sensing Data

Remote sensing data was analyzed by fractal geometry method to quantitatively explain remote sensing images. Three fractal dimension measurement methods, the line-divider (or contour) method, the variogram method and the trianguiar prism method. were programmed. The Landsat TM data of Tengchong area were used to compute the fractal dimensions by the line-divider method. That the contour length was computed by the row method, the column method and the row-column method can find oriented image patterns. When the patterns of images are obviously oriented, measured D values from the row method are different from the column method. Measured D values from the row-column method are hardly influenced by oriented patterns. When R (the coefficient of determination) > 0.9, all D values from them can present the spatial interrelationships within image data. By fractal measurement,both the characteristics of the whole image data-base and each value of the image data-base can be described, the disorder in appearance and the rule in inherence in remote-sensing data can be brought to light,and it will be possible to relate D values to ground objects to explore useful information of various natural phenomena.