Digital Image Processing

Introduction to Supervised Classification:

The purpose of the following assignment serves as a means to introduce Hexagon Geospatial ERDAS Imagine 2014 software. It gives an opportunity to demonstrate the skills and knowledge learned in workshops by creating a subset from an image and performing the methods of both an unsupervised classification as well as a supervised classification. Through both classification methods, it was learned how to adjust parameters and algorithms in order to produce an accurate yet detailed final output map. Furthermore, upon completion of both classification methods, the assignment taught how to read and analyze the statics relating to the images and how they are able to tell thorough information about the area through the digital image. Lastly, the ability to gain practical experience using a remote sensing software and learning additional digital image processing techniques, is valuable when working with satellite imagery.



Principal Component Analysis:

The purpose of the following assignment served as a means to introduce the Principal Component Analysis (PCA) transformation on digital satellite imagery. The knowledge gained from the assignment included learning how to perform a PCA transformation within Hexagon Geospatial’s ERDAS Imagine 2014 and demonstrated the ability to compare the results of the PCA transformation to other classifications types, such as an unsupervised classification, of the same image. The assignment also gave the opportunity to demonstrate the skills and knowledge learned in workshops by adjusting parameters of the PCA transformation to produce an accurate map displaying the land use/land cover (features) of the area being studied. Lastly, the assignment aided in the familiarity and understanding of how to read and analyze the calculated statistics derived from the images in the form of scatterplot graphs, histogram charts, and eigenvalues in order to communicate additional information, in a mathematical form, about the examined area in the image and its attributes.



Geometric Correction, Orthorectification and Mosaicking:

In digital image processing, the use of geometric correction is highly valuable when dealing with problem imagery: “raw digital images usually contain geometric distortions so significant that they cannot be used directly as a map base without subsequent processing” (Lillesand, Kiefer, & Chipman, 2008, p. 486). As a result, geometric correction gives the ability to fix or resolve those distortions so that the image, when corrected (orthorectified), will have the “highest practical geometric integrity” (Lillesand, Kiefer, & Chipman, 2008, p. 486).

To perform geometric correction on problem imagery, a two-step process is used to evaluate the predictability of those distortions. Because aerial imagery can result in several different types of distortions, such as altitude variation, velocity of the sensor, earth curvature, atmospheric refraction, and relief displacement, for example, the Earth’s surface becomes faulty and misrepresented (Lillesand, Kiefer, & Chipman, 2008). Therefore, geometric correction can be implemented to see if those distortions are ‘systematic’ (predictable) or ‘random’ (unpredictable) and depending on the distortion type, how geometric correction is used to compensate those faults.