Linda E. Saavedra, Philip Virgil B. Astillo, Beethoven M. Arellano, Michael E. Loretero, Evelyn B. Taboada, Renante R. Violante, Roland Emerito S. Otadoy
The need to do an inventory of agricultural resources in the Philippines has become a priority due to the increase of natural calamities that hit the country in the recent months. As a start, agricultural resource mapping is done with the main goal of classifying land use and land cover and identifying agricultural features. This paper examines the results of applying multi-resolution segmentation and support vector machine in identifying agricultural features in a large area using the whole municipality of San Fernando as a test case. A varying scale parameter was used to obtain a decent object image. Support Vector Machine (SVM) was implemented to extract agricultural features from the segmented image. Although agricultural features were identified, misclassifications of some objects still exist. To lessen the misclassified objects, SVM was applied following a binary tree approach. An agricultural map showing the extracted land use and land cover was then made and validated. While multi-resolution segmentation and SVM showed promising results, the overall accuracy was only 64%.
Phil-LiDAR Research Center, University of San Carlos, Cebu City, Cebu, 6000, Philippines; Department of Computer Engineering, University of San Carlos, Cebu City, Cebu, 6000, Philippines; Department of Mechanical Engineering, University of San Carlos, Cebu City, Cebu, 6000, Philippines; Department of Chemical Engineering, University of San Carlos, Cebu City, Cebu, 6000, Philippines; Theoretical and Computational Science and Engineering Group, Department of Physics, University of San Carlos, Cebu City, Cebu, 6000, Philippines
