52
J.Jensen
ImagelMap Cartograpby
36. Scaled Postscript Level II output of images and maps
Geograpbic Information Systems (GIS)
37. Raster (image) based GIS
38. Vector (polygon) based GIS (must allow polygon overlay)
Integrated Image Processing and GIS
39. Complete image processing systems (Functions I through 36 plus utilities)
40. Complete image processing systems and GIS (Functions I through 43)
Utilities
41. Network (Internet, local talk, etc.)
42. Image compression (single image, video)
43. Import and export of various file formats
3.2.1 Preprocessing
Remote sensor data must be carefully preprocessed before information can be
extracted from it. The scientist must radiometrically correct the remote sensor data
to remove a) system introduced error (e.g. systematic noise or stripping) and/or b)
environmentally introduced image degradation (e.g. due to atmospheric haze).
Many remote sensing projects do not require detailed radiometric correction.
However, projects dealing with water quality, differentially illuminated mountainous terrain, and subtle differences in aquatic vegetation health and vigor do require careful radiometric correction (e.g. Jensen et al., 1995; Bishop et aI., 1998).
Therefore, it is important that the application software provide a robust suite of
radiometric correction alternatives.
Hydrologists and water management scientists require that most spatial information derived from remote sensor data be reprojected into a standard map projection suitable for input to a GIS. This involves rectification of the remote sensor
data to a Universal Transverse Mercator (UTM) or other map projection using
nearest-neighbor, bilinear interpolation, or cubic convolution resampling logic.
The image processing software should allow ground control points (GCPs) to be
easily and interactively identified on the base map and in the unrectified imagery.
The GCP coordinates are used to compute the coefficients necessary to warp the
unrectified image to a planimetric map projection. The accuracy of the image-tomap rectification or image-to-image registration is specified in root-mean-squareerror (RMSE) units, e.g. the pixels in the image are within ±1O m of their true
planimetric location. The user must also be able to specify the geoid and datum
(e.g. NAD83 refers to the North American Datum 1983 to which all U.S. digital
orthophoto quarterquads must be referenced).
Change detection projects are especially dependent upon accurate geometric
rectification (or registration) of multiple date images. Therefore, it is imperative
that the image processing system software perform accurate geometric rectification.
3.2.2 Display and Enhancement
Digital image processing systems must be able to display individual black and
white images (usually 8-bit) and color composites of three bands at one time (24-
J.Jensen
ImagelMap Cartograpby
36. Scaled Postscript Level II output of images and maps
Geograpbic Information Systems (GIS)
37. Raster (image) based GIS
38. Vector (polygon) based GIS (must allow polygon overlay)
Integrated Image Processing and GIS
39. Complete image processing systems (Functions I through 36 plus utilities)
40. Complete image processing systems and GIS (Functions I through 43)
Utilities
41. Network (Internet, local talk, etc.)
42. Image compression (single image, video)
43. Import and export of various file formats
3.2.1 Preprocessing
Remote sensor data must be carefully preprocessed before information can be
extracted from it. The scientist must radiometrically correct the remote sensor data
to remove a) system introduced error (e.g. systematic noise or stripping) and/or b)
environmentally introduced image degradation (e.g. due to atmospheric haze).
Many remote sensing projects do not require detailed radiometric correction.
However, projects dealing with water quality, differentially illuminated mountainous terrain, and subtle differences in aquatic vegetation health and vigor do require careful radiometric correction (e.g. Jensen et al., 1995; Bishop et aI., 1998).
Therefore, it is important that the application software provide a robust suite of
radiometric correction alternatives.
Hydrologists and water management scientists require that most spatial information derived from remote sensor data be reprojected into a standard map projection suitable for input to a GIS. This involves rectification of the remote sensor
data to a Universal Transverse Mercator (UTM) or other map projection using
nearest-neighbor, bilinear interpolation, or cubic convolution resampling logic.
The image processing software should allow ground control points (GCPs) to be
easily and interactively identified on the base map and in the unrectified imagery.
The GCP coordinates are used to compute the coefficients necessary to warp the
unrectified image to a planimetric map projection. The accuracy of the image-tomap rectification or image-to-image registration is specified in root-mean-squareerror (RMSE) units, e.g. the pixels in the image are within ±1O m of their true
planimetric location. The user must also be able to specify the geoid and datum
(e.g. NAD83 refers to the North American Datum 1983 to which all U.S. digital
orthophoto quarterquads must be referenced).
Change detection projects are especially dependent upon accurate geometric
rectification (or registration) of multiple date images. Therefore, it is imperative
that the image processing system software perform accurate geometric rectification.
3.2.2 Display and Enhancement
Digital image processing systems must be able to display individual black and
white images (usually 8-bit) and color composites of three bands at one time (24-