Sundari, Suci (2017) Identifikasi Kuantitas Bahan Organik Tanah Untuk Tanaman Atsiri Melalui Citra Tanah Menggunakan Metode Jaringan Saraf Tiruan Backpropagation. Sarjana thesis, Universitas Brawijaya.
Abstract
Tanaman atsiri adalah tanaman yang menghasilkan minyak. Minyak tersebut digunakan untuk industri kosmetik, parfum, antiseptic, dan lain-lain. Data statistik ekspor-impor dunia menunjukkan bahwa konsumsi minyak atsiri setiap tahun selalu naik sekitar 10%. Salah satu contoh tanaman atsiri yang terpenting di Indonesia adalah nilam (Pogostemon Cablin Bent). Indonesia menjadi pemasok minyak nilam terbesar di pasar Internasional, yakni sebesar 85%. Terdapat sebuah penelitian yang menunjukkan bahwa produksi minyak nilam lebih tinggi jika tanah mengandung 75% bahan organik+25% NPK (Nitrogen, Phospor, Kalium). Sehingga sangat penting mengetahui kuantitas bahan organik tanah agar dapat diambil tindakan pemupukan yang tepat. Salah satu cara untuk mengetahui kuantitas bahan organik tanah adalah dengan metode analisis laboratorium. Namun metode ini mempunyai kelemahan, yakni bahan kimia yang mahal, memakan banyak waktu, dan proses yang tidak mudah. Penelitian ini bertujuan untuk mempermudah dalam mengidentifikasi kuantitas bahan organik tanah, yakni dengan cara identifikasi melalui citra tanah. Data citra diambil di beberapa kecamatan di Kabupaten Blitar. Terdapat tiga proses yang dilakukan, yakni ekstraksi fitur citra dengan metode color moment dan Gray Level Co-occurrence Matrix (GLCM), serta proses pelatihan dan pengujian dengan metode jaringan saraf tiruan backpropagation. Sistem akan mengidentifikasi kuantitas bahan organik tanah ke dalam lima kelas, yakni sangat rendah, rendah, sedang, tinggi, dan sangat tinggi. Hasil dari penelitian ini didapatkan akurasi tertinggi sebesar 73% dan nilai MSE 0,5122 dengan menggunakan lima fitur GLCM sebagai parameter input, nilai learning rate 0,5, nilai iterasi maksimum 1000, jumlah data latih 210, dan jumlah data uji 12.
English Abstract
Essential crops are plants that produce oil. The oil is used for cosmetic industry, perfume, antiseptic, and others. Statistical data of world import-export shows that the consumption of essential oil always increase about 10% every year. One of the most important examples of essential plant in Indonesia is Patchouli (Pogostemon cablin Bent). Indonesia became the largest supplier of patchouli oil in the international market, which amounted to 85%. The research showing that patchouli oil production will be higher if the soil contains 75% of organic ingredients + 25% NPK (Nitrogen, Phosphorous, Potassium). So it is very important to know the quantity of soil organic matter so that the proper fertilization action can be taken. One way to determine the quantity of soil organic matter is by using laboratory analysis. But this method has a weakness, namely chemicals are expensive, time consuming, and the long process. This study aims to make easier in identifying the quantity of soil organic matter, by using soil image identification. The image data taken in several districts in Blitar. There are three processes are carried out, the image feature extraction method color moment and Gray Level Co-occurrence Matrix (GLCM), as well as training and testing process with backpropagation artificial neural network method. The system will identify the quantity of soil organic matter into five classes, namely very low, low, medium, high, and very high. The results of this study found the highest accuracy is 73% and value of MSE is 0.5122 by using five GLCM features as an input parameter, the learning rate value is 0,5, the maximum iteration value is 1000, with 210 amount of training data, and 12 amount of the test data.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FTIK/2017/72/051701169 |
| Subjects: | 000 Computer science, information and general works > 005 Computer programming, programs, data |
| Divisions: | Fakultas Ilmu Komputer > Teknik Informatika |
| Depositing User: | Kustati |
| Date Deposited: | 09 Feb 2017 15:04 |
| Last Modified: | 21 Oct 2021 01:44 |
| URI: | http://repository.ub.ac.id/id/eprint/147506 |
Preview |
Text
Skripsi_-_125150101111019_-_Suci_Sundari.pdf Download (7MB) | Preview |
Actions (login required)
 |
View Item |