1.7 Resampling and Interpolation

Many image and raster transformations and operations involve pixel or cell resampling and interpolation.

GeoRaster supports the following standard resampling and interpolation methods:

• Nearest neighbor (NN)

• Bilinear interpolation using 4 neighboring cells (BILINEAR)

• Biquadratic interpolation using 9 neighboring cells (BIQUADRATIC)

• Cubic convolution using 16 neighboring cells (CUBIC)

• Average using 4 neighboring cells (AVERAGE4)

• Average using 16 neighboring cells (AVERAGE16)

• OTHER

The keywords for these resampling types are defined in the resamplingType element definition in the GeoRaster XML metadata schema (described in GeoRaster Metadata XML Schema). Except for OTHER, the keywords can be used in several subprograms including the following:

• SDO_GEOR.generatePyramid

• SDO_GEOR.scaleCopy

• SDO_GEOR.reproject

• SDO_GEOR.rectify

• SDO_GEOR_AGGR.append

• SDO_GEOR_AGGR.getMosaicSubset

• SDO_GEOR_AGGR.mosaicSubset

The resampling type OTHER is used only to indicate an unknown or external resampling type when the pyramids of a GeoRaster object are generated or imported from external sources, such as a file.

Raster data deals with real world phenomena that vary continuously over space. This data is usually associated with grid interpolation, a method for interpolating values at spatial positions between the cells or within the cells. In GeoRaster, SDO_GEOR.evaluateDouble is the grid interpolation function. It uses the same keywords for interpolation methods as those for resampling.