| Comparing spectral-based and object-based classification for automatic mapping: A case study, basement rock of Umm-Grayet area |
| Paper ID : 1084-ICRSSSA-FULL |
| Authors |
|
safaa Mohamed Hassan *1, Mohamed R Metwalli2 1NARSS 2Digital image processing and its applications department (DIP), Data Reception, Analysis and Receiving Station Affairs division, National Authority for Remote Sensing and Space Sciences, Cairo 11769, Egypt; |
| Abstract |
| The study area at the southern eastern desert of Egypt is occupied by the Precambrian rocks of the Pan African assemblages including the metavolcanics (metabasalt, meta-andesite and metarhyolite, their equivalent metapyroclastics). This paper compares the results of object-based classification algorithms to a supervised pixel-based classification for mapping basement lithological units in Wadi El Alaqui, Eastern Desert of Egypt. The object-based method involved the segmentation process of the image into objects at several scale levels. The proposed training objects and the support Vector Machine (SVM), Decision Tree (DT) , Random Trees(RT), Bayes and K Nearest Neighbor (KNN) supervised classification algorithms have been used to assign the classes of the lithological units. The pixel-based classification involved the selection of training sets and a classification using the Support Vector Machine (SVM) classifier algorithm. The accuracy assessment using cross-validation and grid search technique of all classifications results were calculated. A comparison of the results shows a statistically significant higher overall accuracy of the object-based classification over the pixel-based classification. The integration of both ASTER and ASTER-MNF layers and associated class rules into the object-based classification produced higher accuracies overall reach to 94.41%. The results indicate object-based analysis has good achievement for automatic extracting the basement lithological units information from ASTER object-based classified image captured over Arabian Nubian belt. |
| Keywords |
| basement rocks, Machine learning classification and automatic mapping |
| Status: Accepted (Oral Presentation) |
