泰国THEOS遥感影像质量评价

选取泰国THEOS(Thailand earth observation satellite)卫星影像作为研究对象,提出从图谱成像质量和专题地类提取性能评价影像应用潜能。采用不同地类的像元均值表征影像的光谱质量;选取灰度共生矩阵GLCM(Grey level co-occurrence matrix)计算得到的角二阶矩ASM(angular second moment)、相关性(correlation)、对比度(contrast)和熵(entropy)评价影像的图像质量;利用面向对象的分类方法评价影像的地物解析能力。结果表明:(1)耕地、城市和林地的均值分别为121.14、73.70和137.62,说明THEOS影像对不同地物有较好的区分性能;(2)对比度和熵的均值最大(分别为18.93和2.88),可以用于表征THEOS影像丰富的纹理信息;(3)2 m分辨率THEOS融合影像可以较好地提取均一地类,但解析多种地类混合的区域效果较差。本研究可为THEOS卫星数据在我国的广泛应用提供参考。

Quality evaluation of the THEOS satellite imagery using textural and spectral features

The THEOS (Thailand earth observation satellite) image quality was evaluated through the spectral and texture quality for imaging and the capability for identifying thematic land-cover types of three methods. Firstly, the mean of pixel values were used to assess the spectral discrimination. Secondly, the ASM (Angular second moment), correlation, contrast and entropy, derived from the GLCM (Grey level co-occurrence matrix), were used to represent the texture features of imagery. Finally, the object-oriented classification method was used to identify the thematic land-cover types. The analysis results were as follow. (1) The mean grey values of cropland, built-up and forestland were 121.14, 73.70 and 137.62, respectively, which indicated that THEOS imagery has good capability in identification of different land cover types; (2) The contrast and entropy (their means were 18.93 and 2.88, respectively) were better to show the texture information; (3) The results were good for homogenous land covers, but poor for the regions with the mixed types of land covers. This study provides a good reference for a wide utilization of THEOS in China.