Levels of Digital Image Processing |Spatial Domain Enhancement | M Tech PhD |

Levels of Digital Image Processing
Low-level image processing covers the primitive operations for
noise suppression, contrast enhancement and image sharpening. In this
level, both input and output are images. Mid-level image processing
refers to image segmentation, representation / description techniques,
using which any given image can be reduced into a form suitable for
object recognition or classification. The input for a mid-level process is
generally images and the outputs are the attributes extracted from those
images such as edges, contours and objects. The high-level image
processing tries to replicate human cognition and ability to make
decisions. Computer vision and image analysis are the illustrations of
high-level image processing.

Basics of Image Enhancement
Irrespective of the application domains of digital image
processing, image enhancement techniques find greater usage in the
pre-processing phase. These techniques alter the information content of
the input image in order to improve its visual impact of the entire image
or a portion / feature of the image.
As the effectiveness of the enhancement techniques are primarily
determined by human visual perception, there is no generic rule to
determine their suitability. However, quantitative metrics are
alternatively used to determine their appropriateness with respect to
their applications.

Spatial Domain Enhancement Techniques
This approach handles any input image represented in two dimensional
space, such that each pixel has unique coordinate pairs and
an intensity value that lies within the predefined dynamic range. Some
techniques use the statistics of the entire image for enhancement,
known as global enhancement, whereas the local enhancement methods
use the statistics of sub-images of the input image. The latter one uses a
window that partitions the input image into uniform-sized sub-images
in the specified order, say n×n (normally n is an odd number). Each
pixel, aligned to the center of the window is manipulated with respect
to its neighboring pixels bounded by that window. Some of the most
commonly used spatial domain methods are
i) Gray Scale Manipulation
ii) Histogram Equalization
iii) Image Smoothing
iv) Image Sharpening
v) High Boost Filtering

Leave a Reply

Your email address will not be published. Required fields are marked *