to degraded accuracy of classification if the existing methods are applied. The
balanced selection depends on the classification tasks.
2. Preprocessing: The preprocessing includes geometric correction, radiometric
correction, and cloud masking. Geometric correction is the process of digitally
removing the geometric distortions to match a specific projected surface or shape
in digital image processing (JARS 1999; Toutin 2004; Mather and Koch 2011).
Geometric distortions are mainly coming from three sources: sensor (observer),
target (observed), and projection (reference) (JARS 1999; Toutin 2004; Mather
and Koch 2011). Most sensor-related distortions (variations in movement, attitude, mechanics, viewing angles, and clock) can be systematically corrected.
These are often done by data provider or satellite sensor-receiving stations. The
remaining distortions from the sensor (i.e., degrading, nonsystematic variations),
Fig. 10.1 General
workflow for crop pattern
mapping with remote
sensing
178
E. G. Yu and Z. Yang
balanced selection depends on the classification tasks.
2. Preprocessing: The preprocessing includes geometric correction, radiometric
correction, and cloud masking. Geometric correction is the process of digitally
removing the geometric distortions to match a specific projected surface or shape
in digital image processing (JARS 1999; Toutin 2004; Mather and Koch 2011).
Geometric distortions are mainly coming from three sources: sensor (observer),
target (observed), and projection (reference) (JARS 1999; Toutin 2004; Mather
and Koch 2011). Most sensor-related distortions (variations in movement, attitude, mechanics, viewing angles, and clock) can be systematically corrected.
These are often done by data provider or satellite sensor-receiving stations. The
remaining distortions from the sensor (i.e., degrading, nonsystematic variations),
Fig. 10.1 General
workflow for crop pattern
mapping with remote
sensing
178
E. G. Yu and Z. Yang