尺度转换问题多年来一直是地理学者研究的热点和难点。但是由于其过程的复杂性、影响因素的多样性、变异性等，很难真正的把握其规律。随着人们对尺度转换问题研究的日益深入，各种各样的方法和理论开始应用于尺度转换。尽管这些尺度转换的方法都有各自的优点，但其应用于遥感尺度转换的局限性和内在不足必然存在。而新的理论和尺度转换方法则需要理论和实用的进一步完善。本文即针对此种情况，一方面对目前常用的一些尺度转换方法进行分析，另一方面侧重检验和完善一些新的尺度转换方法，如直方变差图方法和傅里叶变换方法。直方变差图方法是由张颢等人提出的一种新的关于空间异质性的表达方式。本文在此基础上进一步研究尺度转折点的存在。同时，通过对直方变差图中边界点的研究，找到了直方变差图中聚合窗尺度与图像的灰度变化快慢（图像频率）之间的关系，为尺度转换提供了一个新思路。傅立叶变换主要是通过滤去一副影像中的高频信息来实现从高分辨率图像到低分辨率图像的尺度转换。但是同一地区的不同低分辨率的图像所包含的高频信息是不尽相同的，如何定量的从高分辨率影像转换得到某一低分辨率的近似影像是研究的一个重点。本文假设将遥感过程理解为对连续的地面离散化观测的过程，因此可以利用傅立叶变换中产生的频谱混叠效应来解释不同低分辨率遥感影像所包含的高频信息不同的原因，进而利用此原理确定尺度转换的滤波参数，从而定量地得到各低分辨率的近似影像。试验结果表明取得了较好的效果。最后，将此确定滤波参数的方法应用于地表时空多变要素—叶面积指数（LAI）的尺度转换。通过对ASTER反演的LAI进行尺度转换，得到1KM分辨率的结果，与其他方法的结果和实际观测数据比较，结果表明效果令人满意。

Scale and Scaling problem have been discussed heatedly by geographers for many years. However, it is difficult for people to grasp its laws, because of the complexity of its process and the diversity and the variability of the factors which influence it. In recent years, as the research on scaling is increasingly furthered and developed, more and more methods and theory are applying in the scaling. Although the methods all have their advantages, the limits of them for applying in scaling of remote sensing exist. And the new theory and methods still need to be improved in theory and practice. Aim to this, scaling methods those often used at present were analized at first in this paper. At the other hand, some new scaling methods are validated and improved, such as the Histo-Variogram method and the Fourier transformation method.The Histo-Variogram method was put forward by Zhang Hao etc. It is a new way to express the spatial heterogeneity. In this paper, whether the turning point of scale exists which proposed by Liu Jinbao etc. is further studied. At the same time, fmn-contour in Histo-Variogram is investigated, the relationship between scale and the change speed of image are found. This may provide a new mind of the scaling problem.According to filtrate the information of the high-frequency of image, Fourier transformation can achieve scaling-up transformation. But Different low-resolution image have different information of the high-frequency. So it is important to study how we can get a particular low-resolution image. In my paper, I consider that the process of remote sensing is the discretization in sequence ground. So we can use the effect of Aliasing to explain why different low-resolution image have different high-frequency information. Base on this theory we can select the advisable filter coefficients to get the special low-resolution image. I also extend this method to the transformation of LAI. I transformate the LAI which was inverted through the data of ASTER to 1KM resolution. In order to verify the result of experiment we compare the result with the other results which get from the other methods and the real value of LAI. It make clear that the filter coefficients is effective.