地理国情普查中地表覆盖自动解译问题研究

Research on the Automatic Interpretation of Surface Coverage in Geographical Conditions Census

针对地表覆盖自动解译时经常遇到的分割尺度设置、影像噪声、判读差异、DLG数据与影像不一致等问题,在实际应用中,根据影像分辨率与测区地类设定合适的分割尺度,采用合理的解译顺序、调整分割尺度和采样率来降低噪声影响,用人机交互解译来解决判读差异问题,人工解译与现场核实相结合判断DLG数据与影像不一致区域。以上措施较好地解决了自动解译遇到的问题。

Earth surface coverage automatic interpretation often encountered the problems of segmentation scale settings, image noise, interpretation differences and the differences between DLG data and images and other issues. In practical applications, we set the appropriate segmentation scale based on image resolution and measurement area class, and used reasonable order of interpretation, adjusting the scale division and sample rate to reduce the effects of noise. We used human interaction interpretation to resolve the problem of differences in interpretation, and used combination of artificial interpretation and field verification to determine the areas where DLG data were inconsistent with the images.