Page 35 - Digital Analysis of Remotely Sensed Imagery

8 Cha pte r O n e

1.3.1 Data Preparation
Core to data preparation is image preprocessing. Its objective is
to correct geometrically distorted and radiometrically degraded
images to create a more faithful representation of the original
scene. Preprocessing tasks include image restoration, geometric
rectification, radiometric correction, and noise removal or sup-
pression. Some of these tasks may have been performed at a ground-
receiving station when the data are initially received from the
satellite. More preprocessing specific to the needs of a particular
project or a particular geographic area may still be performed by the
image analyst.

1.3.2 Image Enhancement
Image enhancement refers to computer operations aimed specifically at
increasing the spectral visibility of ground features of interest through
manipulation of their pixel values in the original image. On the
enhanced image it is very easy to perceive these objects thanks to their
enhanced distinctiveness. Image enhancement may serve as a
preparatory step for subsequent machine analysis such as for the
selection of training samples in supervised classification, or be an end
in itself (e.g., for visual interpretation). The quality or appearance of an
image can be enhanced via many processing techniques, the most
common ones being contrast enhancement, image transformation, and
multiple band manipulation.

1.3.3 Image Classification
Image classification is a process during which pixels in an image are
categorized into several classes of ground cover based on the
application of statistical decision rules in the multispectral domain or
logical decision rules in the spatial domain. Image classification in the
spectral domain is known as pattern recognition in which the decision
rules are based solely on the spectral values of the remote sensing
data. In spatial pattern recognition, the decision rules are based on
the geometric shape, size, texture, and patterns of pixels or objects
derived from them over a prescribed neighborhood. This book is
devoted heavily to image classification in the multispectral domain.
Use of additional image elements in performing image classification
in the spatial domain is covered extensively, as well, together with
image classification based on machine learning.

1.3.4 Accuracy Assessment
The product of image classification is land cover maps. Their accuracy
needs to be assessed so that the ultimate user is made aware of the
potential problems associated with their use. Accuracy assessment is
a quality assurance step in which classification results are compared
with what is there on ground at the time of imaging or something that