Home > Thesis > Mahesh R. Huchhe


Page 16 of
| Previous | Next |

| TOC |

Input Data and Methodology



3.4.1 Image Processing    
DIP was carried out in the computer system, which has the following hardware & software configuration.

  • 80 GB hard disk, Window XP version Pentium(R) IV processor.
  • ERDAS Image processing 8.6 and
  • ArcGIS 9.1 version.

Geometric correction of Image and Preparation of Area of Interest (AOI)
A first order polynomial transformation and resampling with nearest neighborhood algorithm was used to spatially geo-reference the IRS ID P6 LISS III image to projection System, Ground control points identifiable at road intersections in reference to topographic map (1:250,000) from SOI toposheet were used. The boundary of the study area was extracted using Area of Interest (AOI) Module of the ERDAS 8.6. This AOI of the study area was then used in extracting Nagpur district sub-image. The procedure followed is presented schematically in Fig. 3.


Digital supervised and unsupervised classification
Supervised classification is defined as the process of samples of pixels of known identity to classify pixels of unknown identity. Samples of known identity are those pixels located within training areas. Pixels located within these areas called as the training samples are used to guide the classification algorithm to assign specific spectral values to appropriate informational class. The basic steps involved in a typical supervised classification procedure are:

  • FCC Interpretation.
  • The training stage.
  • Feature section.
  • Selection of appropriate classification algorithm.
  • Post classification smothering.
  • Accuracy assessment.

The training sites generated in the present study include current fallow, road, river, forest, rock out crop and water body. Schematic diagram of supervised classification is also presented in Fig. 3.

Unsupervised classification is a process of grouping pixels that have similar spectral values. Unsupervised classification procedures generally require no knowledge of existing cover type prior to classification. Classes that result from unsupervised classification are spectral classes because they are based solely on the natural grouping in the image values. The identity of the spectral class will not be initially known. The analysis must compare the classified data with some of reference data to determine the identity and informational value of the spectral classes.

| Previous | Next |