Digital Image Processing Print E-mail

Digital image processing (DIP) involves manipulation of a digital image by a computer to aid information extraction and further analysis. For technical details on the specific DIP algorithms and how the computer implements them the readers are referred to the resources outlined in the Further Reading subsection. Here only examples of processing results from commonly used DIP techniques are presented.

The most common DIP operations on single band images involve contrast enhancements and image filtering (smoothing and sharpening). Contrast enhancement is an important step in improved visualization of digital image data.

Contrast enhancing an image (courtesy A. Prakash)
Contrast enhancement techniques often take advantage of the full range of pixel values (digital numbers) possible for the image

Image filtering is carried out to either smooth an image (suppress unwanted noise) or to sharpen the image (enhance roads, faults, shore lines, etc). In image filtering each pixel in the output image is computed as a function of one or several pixels in the original image. Sophisticated image filtering operations can also help to enhance features in a pre-selected direction. For example on an image of a structurally complex terrain a user can choose to preferentially enhance all lineaments (faults, joints, fold axes, etc) that trend in the northeast – southwest direction.

Image smoothing (courtesy A. Prakash)
Image smoothing suppresses the high frequency changes (abrupt changes in DN values) and enhances the low frequency changes (smooth changes in DN values)

Image sharpening (courtesy A. Prakash)
Image sharpening enhances the high frequency changes and reduces the low frequency changes. On a sharpened image the difference between the DN values of adjacent pixels becomes larger.

Operations on multiband images are more sophisticated. One can generate color images from three grey scale images. To generate color images the three grey scale images need to be displayed in the three primary colors red, green and blue. A combination of different proportions of the three primary colors gives the full spectrum of colors.

When images acquired in the red, green and blue parts of the electromagnetic spectrum are displayed in red, green and blue color, respectively, the output composite image is called a true or natural color image. On a true color image, the colors of an object are similar to the colors the human eye perceives in real life – i.e., a green tree appears green, a blue lake appears blue, etc.

Any image acquired in any part of the EM spectrum can be coded in any color to generate a color composite image. However, the colors in the output image will vary depending on the choice of the images and the choice of the color in which they are displayed. As the colors of objects on the output image are different from what a human eye perceives, these images are called false color composites.

Color composite images (courtesy A. Prakash)
On a false color composite, colors appear different from what the human eye perceives. Great care needs to be exercised in interpreting the meaning of the colors on a false color composite

See also the Remote Sensing learning module, the Polar Remote Sensing module and the Land Use / Land Cover Change learning module in the Exemplary Learning Module section.

Digital image processing also relies on more advanced mathematical and statistical operations on the digital image to enhance the features of interest. Following are some student projects that exemplify higher order DIP techniques such as PCA, IHS, and CRC.
 

Stratigraphic and Structural Mapping, Eastern Brooks Range, Alaska (Rebecca Bailey

Landcover classification of the Toklat Basin Denali National Park using Landsat ETM data (Larissa Yocum)

Comparing the geological map of Northern foothills of the Alaska Range with signatures seen in Landsat imagery (Cheryl Robar)
 
< Prev   Next >
Joomla Templates by Joomlashack
Joomla Templates and Joomla Tutorial