【摘要】 高分辨率图像景观空间结构和特征尺度的分析是基于图像处理获取地物详细信息的基础。本文以泉州市为例,对不同类型景观区域图像(城市景观,农田景观,森林景观,道路和绿化带景观等)进行了基于小波分解的不同尺度小波系数地统计分析和原图像的半方差分析。研究分析表明,基于小波分析的不同尺度小波系数标准差随其变化的空间分辨率的变化可以概要地表征地物的景观特征尺度,小波分析作为一种景观格局研究方法是可行的。半方差分析能将试验区景观结构的最主要总体平均结构特征反映出来,但对不同方向结构特征的分析并非像小波分析那样敏感,如能利用已有研究区的先验知识,分析结果可能会更好些。更多还原

【Abstract】 Understanding the spatial structure of fine spatial resolution images is instrumental for either pixel-or object-based image analysis.In this paper,the characteristic scale of scene variation in images is evaluated using statistics of sub-images produced by a wavelet transform And the semivariances of the images is calculated.As an example,the study analyze the city landscape,farmland landscape,forest landscape and road and green belt landscape pattern on the Quickbird image(spatial resolution 2.4m).It was found that with energy signature images,the change rate of SD over spatial resolution range between two successive decomposition levels(ΔSD/ΔR) suggested a synoptic and approximate description for the characteristic scale of scene variance.It is feasible that to evaluate the scene variation using wavelet-based methodology.The semivariance analysis can detect the landscape structure overall average pattern on the experimental data,but it is not as sensitive as wavelet-base methodology to directional structure characteristics scale analysis.The semivariance would do well when the more experience or knowledge about the study area was input.更多还原