Information Content

• Published in: Scale in Spatial Information and Analysis pp. 241 - 278
• Main Authors: Zhang, Jingxiong, Atkinson, Peter, Goodchild, Michael F.
• Format: Book Chapter
• Language: English
• Published: United States CRC Press 2014; Taylor & Francis Group
• Subjects: ENVIRONMENTAL ENGINEERING & TECHNOLOGY; Geographical information systems (GIS) & remote sensing; Human geography; NL NL NL; Hyperspectral Image Data; Spatial Dependence; Random Variable; Compressed Sensing; Entropy Rate; Mutual Information; Continuous Random Variables; Radar Images; Discrete Random Variable; PWS; NL NL; Radiometric Resolution; Multiplicative Noise; Differential Entropy; Information Theory; Super-resolution Mapping; Voronoi Regions; Information Content; Relative Entropy; Raster Maps; Conditional Entropy; Area Class Maps; Joint Entropy
• ISBN: 9781439829370; 1439829373
• DOI: 10.1201/b16751-15

Remotely sensed images, both optical and microwave (radar), have become major sources of spatial information. As described in Chapter 3, these images are formed by recording the reflected radiance or energy from a scene. A typical digital image consists of a two-dimensional array of pixels, each representing the average reflectance, emittance, or backscattering of the surface within the sensor’s instantaneous field of view, as described in Chapter 3. The images are used to detect the presence of certain phenomena, map their spatial extents, and estimate certain biophysical variables, such as leaf-area index and surface temperature. Usually, the raw images are processed using various operations, such as filtering, compression, enhancement, transformation, feature extraction, thematic classification, and others, where the analyst seeks to maximize the information content in the images for certain applications (Lee 1980; Bell 1988; Price 1994; Price 1997; O’Sullivan et al. 1998; Narayanan et al. 2002; Peng et al. 2005). The information extracted from images is usually depicted in the form of raster or vector maps. It is important to be able to quantify amounts of information provided through the processes of measurement, geo-processing, and representation of spatial entities and distributions, so that we can increase the information potential of a dataset for a particular application (Barnsley et al. 1997; Goodchild 2003; Harrie and Stigmar 2010).