The information content of remote sensing imagery depends upon various factors such as spatial and radiometric resolutions, spatial scale of the features to be imaged, radiometric contrast between different target types, and also the final application for which the imagery has been acquired. Various textural measures are used to characterize the image information content, based upon which different image processing algorithms are employed to enhance this quantity. Previous work in this area has resulted in three different approaches for quantifying image information content, primarily based on interpretability, mutual information, and entropy. These approaches, although well refined, are difficult to apply to all types of remote sensing imagery. Our approach to quantifying image information content is based upon classification accuracy. We propose an exponential model for information content based upon target-background contrast, and target size relative to pixel size. The model is seen to be applicable for relating information content to spatial resolution for real Landsat Thematic Mapper ( TM) as well as Shuttle Imaging Radar-C ( SIR-C) images. An interesting conclusion that emerges from this model is that although the TM image has higher information content than the SIR-C image at smaller pixel sizes, the opposite is true at larger pixel sizes. The transition occurs at a pixel size of about 720 m. This tells us that for applications that require high resolution (or smaller pixel sizes), the TM sensor is more useful for terrain classification, while for applications involving lower resolutions (or larger pixel sizes), the SIR-C sensor has an advantage. Thus, the model is useful in comparing different sensor types for different applications. References: 22
展开▼
