| The overall objective of this research study was to investigate one of classification problems and to improve classification accuracy, by removing electromagnetic radiation scattering effect and incorporating Global Positioning System (GPS) ground truth data with remote sensing data. This research investigated the possibility of improving classification accuracy by simultaneously considering eliminating the problem of atmospheric scattering of Mie particles.; Remote sensing data represented by four scenes in different dates for two different locations; Milwaukee and Nicolet in Wisconsin has been used.; Each experiment is done to derive the unknown constant of the size distribution using training sets throughout the image for iterations including atmospheric scattering correction; pre-classification and pre-evaluation are employed for this stage. Post-processing stage includes classification and evaluation using independent training sets, GPS ground truth data, and KHAT statistics.; The research result of corrected images showed great improvement of image quality such as color, features, tone, and contrast as it is compared with original images. The classification accuracy improved significantly between 25% to 11% as overall accuracy and Z-score is between 11.91 and 4.0 which was done through KHAT statistics evaluation. |
