Information capacity is a grayscale digital image quality evaluation model based on multi-dimensional histogram. As local region characteristics of pixels are taken into account in the calculation, it can objectively and effectively characterize the structural feature information of these ground objects. Information capacity is related to image gray levels, while gray levels of remote sensing imagery reflect the complexity of surface landscape. Information capacity is firstly used in geo-science research in this paper, and spatial variation of information capacity in different landform areas is discussed. The results show that spatial the variation of information capacity is closely related to complexity of regional surface structure in different landform types. Generally, information capacity of mountain is the largest, information capacity of hills and plain is followed in order. Moreover, information capacity is sensitive to the change rate of vegetation coverage of different land covers. There are strong correlation between information capacity and NDVI standard deviation, and the correlation coefficient respectively is 0.8347 or 0.8648 in different experimental areas. This study shows that information capacity can effectively characterize the complexity of the regional surface structure and have great significance for quantitative research of surface feature complexity.