Classifying Spectral Signatures In Multi-spectral Imagery

 

Landcover Classifying 

Today's exercise was all about classifying multispectral imagery based on the pixel values of each spectral signature or layer in our image. An example would be trees, water, buildings etc... all getting assigned a certain value to be recoded in our image classification. 

Original Image of Natural Color

Same Image Color Infared 

By classifying our image, it becomes a thematic raster file. We can "color code" our imagery to determine what types of land use exists in our image. 

From there, we can calculate the total acreage of land use for trees, grass, buildings (urban development) and even the road ways in a given area. This is extremely powerful to provide to local government policy makers in determining different decision making techniques when deciding on urban development, deforestation, building new roads, protecting environmental waterways and seeing how much of an area ca change overtime. 

Land Cover Analysis - Supervised Classified Raster

Now we know a good idea of how much land types we have in this city area, as well as acreage by land type. We used Erdas Imagine to classify this raster using a supervised classification method, and then recoded our image to identify the correct spectral signatures on our image. We then applied a color scheme that made each type of land stand out visually and  calculate the geometry of area for each land type.