TM影像解译中去除云层及其阴影方法研究

A study on the technology of removing clouds and their shadows during TM image interpretation

目的对TM(Thematic Mapper,专题制图仪)影像上的云层及其阴影进行处理以识别被覆盖的地物。方法在遥感图像处理软件ERDAS的支持下,综合运用该软件提供的多种图像处理方法,分别进行去除云层及其阴影试验。结果利用空间卷积运算、傅立叶变换、傅立叶图像编辑、傅立叶逆变换组合和缨帽变换的方法有效地去掉了试验区图像上的云层;利用去霾处理和主成分分析方法按一定顺序组合将有效地去除云层的阴影。结论该方法及其思想不仅可用于去除云层和阴影对遥感图像质量的影响,而且也可用于其他图像处理中,如影像的割补、分类后处理等。

Aim To detect the covered elementary composition and features by dealing with the cloud and its shadow of TM image. Methods With the support of ERDAS, the cloud and its shadow were removed individualy through multi - image processing methods provided by ERDAS software. Results The clouds can be effectively removed by applying the Convolution Operation, Fourier Transform, Fourer Transform Editor, Inverse Fourier Transform and Tasseled Cap; The cloud shadow can be removed by applying the Haze Reduction and Principal Components Analysis. Conclusion These methods and ideas can be used not only to remove the influence of cloud and its shadow on the quality of remote image but also to process other images, such as patching and recoding images.