Topic 05 Flashcards
(25 cards)
what are geometric transformations
correcting nonsystematic issues
often problems with imagery
one coordinate system to another
raster/vector
systemic distortions
predictable errors due to nature of data capture (eg satelitte sensor)
usually easily corrected
random distortions
distoritaions that are due to unforseen changes in the sensors geometry
corrected by using ground control points(GCPS) and regression techniques
map transformations
transforming digilitzed features or imagery into a particular spatial reference system
control points
estbalished known points
georeferencing
analyzing individual locations to locate features
transformation equation
a mathematical equation that is used to match control points to corresponding locations in a projected oordiante system
dufferent levels of mathematical equations
equiarea
similarily
affine
projective
cannot be too exact with control points
affine transformation
allow for rotation, translation, skew, and differential scaling while preserving line parallelism
changing the resolution (cell size)
decrease amount of oricess in your storage
to make another raster have the same cell values so they can be compared
resampling (changing raster cells)
smaller pixel to larger pixel (losing value/information)
cannot go from a larger pixel to a smaller pixel
georeferencing raster images
transforming raster involves moving from immage space to coordinate space
raster registration
alignment of two rasters
georeferencing definition
attaching geographic coordinates to each point in a raster
rectification
removing distortions caused by dara capture process
GCPs to known map locations
orthorectification
removal of distortions caused by terrain relief
rectification steps
- spatial interpolation (rubber sheeting)
- Digital number interpolation (resampling)
raster resampling
issue with raster in re scaling
cubic convolution
(transitions look better but destroy statistic structure)
transformation and resampling data always involves losing some data quality
raster resampling techniuqes (4)
basic types (in order of increasing computation)
nearest neighbour
majority rule
bilinear interpolation
cubic convolution
can you chain transformations together
no always use the original
nearest neighbour
uses the pixel value from the nearest neighbor
maintains data integrity of the image
bilinear interpolation
distance-weighted average of 4 neasrest neighbours
alters the original pixel values
cubic convolution
uses teh 16 nearrest neighbours
longer processing times
smoother images
root mean square error (RMSE)
with interval/ratio data, accuracy is often described with a root mean square error
standard deviation
deviation between the input and the output