A spectral and textural analysis of digitized high altitude color infrared photography of residential areas in Aurora, Illinois

Bibliography: pages 69-70.The numerical analysis of remotely sensed data is most often associated with the classification of land use and land cover. These techniques are well suited for the analysis of spectrally differentiable land use or land cover areas; however, no distinction is possible between diverse land use or land cover having spectrally similar occurrences. Photointerpreters form conclusions about an image's content based on several criteria. One criterion is the spatial arrangement of tone or color. Another criterion is spatial perception which is further divided into context, the interrelationship of image features, and the characteristic which is the focus of this study, texture. Texture is broadly defined as the distribution of tonal or color differences. The quantitative description of these distributions should yield an additional criterion helpful for differentiating spectrally similar, spatially differentiable, land cover categories. The above contention was tested by generating measures of spatial variability. The textural "signatures" of small image segments were used in conjunction with spectral data in an attempt to differentiate spectrally similar land use/land cover categories. Residential areas were examined because of their spectral similarity to other land cover categories such as agricultural and open land. In general, the spatial information did aid in the differentiation of residential areas from surrounding vegetated land use/land cover categories as well as from other urban areas.M.S. (Master of Science