| dc.description.abstract | Penelitian ini mengkaji perbandingan teknik klasifikasi berbasis objek untuk mendeteksi penutupan tajuk hutan lahan kering di PT. Inocin Abadi, Kabupaten Boven Digoel, antara algoritma watershed dan mean-shift. Algoritma watershed menghasilkan nilai akurasi yang lebih tinggi dibandingkan algoritma mean-shift, tetapi dalam waktu pemrosesan algoritma mean-shift lebih efisien daripada algoritma watershed. Algoritma watershed menghasilkan nilai KA terbesar pada algoritma watershed 70,83% dan OA sebesar 94% pada kombinasi parameter (5-5-20) dengan waktu 1 jam dan 2 menit, lalu algoritma mean-shift memiliki nila KA terbesar 69,32% dan OA sebesar 94,22% pada kombinasi parameter (5-20-10) dengan waktu 3 menit 40 detik.
This study examined two algorithms of the object-based classification techniques, i.e., namely the watershed and the mean shift, to detect dryland forest canopy cover at PT. Inocin Abadi, Boven Digoel Regency. The study shows that the watershed algorithm provided a higher accuracy value than the mean-shift algorithm; however, the processing time of the mean-shift algorithm is more efficient than the watershed algorithm. The watershed algorithm shows a KA value of 70.83% and OA of 94% with the parameter combination of 5-5-20 with a processing time of 1 hour and 2 minutes, while the mean-shift algorithm provided a KA value of 69.32% and OA of 94.22% with the combination of parameters 5-20-10 with a processing time of 3 minutes and 40 seconds. | id |
