ERSC 3P07 Lecture Notes - Lecture 6: Soil Classification, Rangeland, Categorical Variable
Document Summary
Spectral transformations of data: mapping and monitoring vegetation, vegetation indices, image classification. Spectral transformations: used to modify pixel brightness values based solely on each pixel"s value (aka: a point operator). Spatial transformations: used to modify a pixel"s brightness value based on the values of the surrounding (neighbourhood) pixels (aka: a local operator). Spectral transforms alter spectral (or feature) space and ignore spatial information. Land-use and land-cover classification of south dumfries township, ontario. To extract thematic information (categorical data) from data. To place a landscape into categories of land-use and land-cover (e. g. , forest, agriculture, urban, rural, wetland, water, etc. ) An automated procedure used to extract thematic information from remotely sensed data. Note: focusing on hard classification types (one class/pixel) Unsupervised: statistical clustering algorithms used to select spectral classes inherent to the data. Supervised: the image analyst supervises the selection of information classes that represent patterns or land-use/land-cover features that the analyst recognizes. x-y scatterplots (multispectral feature space)