The purpose of this study was to classify the use of multiple methods of expert systems, to determine the optimal model of udearly detection of illegal logging and changes in a forest area. Illegal logging is one of the ecological concerns related to the udforest protection in Indonesia. In some cases, the illegal logging is almost impossible to be early identified by only physical udinspection from the forestry staffs due to the field difficulty to reach the location. For this reason, the image processing by udmeans Expert System using Support Vector Machine (SVM) method might be helpful to early identify the illegal logging udactivity. The sample image used in this research is taken from the Google Earth for one location of river basin area in udIndonesia with the size area of 560m x 375m. The procedure of identification system is started by cropping the digital image udand then classified for retrieving two-halves of the image in the form of forest and non-forest. These forms will be the input udfor the support vector machine method. Meanwhile, the testing data consists of five pairs of initial and final data. The udsimulation result shows that SVM method is able to early identify of illegal logging activity in protected forest areas with a udlevel of accuracy of 80%. The testing process is basically similar to the image processing; only process of saving data sets it udapart. The main point is the illegal logging indicator is obtained through the segmentation stage. Further analysis may uddetermine the output whether the area is illegal logging or not.
展开▼
