Sedo, Raimundus and Dr. Eng. Panca Mudjirahardjo,, S.T., M.T. and Dr. Ir. Erni Yudaningtyas,, M.T. (2019) Identifikasi Takaran Pupuk Nitrogen Berdasarkan Tingkat Kehijauan Daun Tanaman Padi Menggunakan Metode Histogram of s-RGB dan Fuzzy Logic. Magister thesis, Universitas Brawijaya.
Abstract
Analisis warna daun padi ialah suatu cara mengidenfikasi kandungan unsur hara yang perlu dilakukan sebagai dasar rekomendasi pemberian pupuk pada padi. Jika tanman padi kelebihan nitrogen, maka akan gampang terserang penyakit dan mencemarkan air tanah. Sebaliknya, bila kekurangan nitrogen, maka pertumbuhannya akan terhambat. Penelitian ini bertujuan untuk merancang sistem identifikasi untuk analisis takaran pupuk nitrogen sesuai dengan tingkat kehijauan daun padi melalui konsep pengolahan citra dengan metode Histogram of s-RGB dan Fuzzy Logic berbasis android. Pada penelitian ini, Bagan Warna Daun (BWD) sebagai konsep dasar pada proses pengembangan sistem ini. Sistem didesain menurut 4 skala sesuai level warna BWD sehingga dapat menganalisis citra daun padi sebagai dasar rekomendasi takaran pupuk nitrogen yang diperlukan tanaman padi. Pengujian dilakukan pada dua buah smartphone dengan kapasitas resolusi kamera yang berbeda, yaitu smartphone 8 MP dan 5 MP. Berdasarkan hasil pengujian menggunakan metode Euclidean Distance, diketahui jarak terdekat rata-rata nilai RGB sistem terhadap nilai RGB pada BWD sekitar 12,61 untuk smartphone 8 MP, sementara smartphone 5 MP mencapai 13,97. Evaluasi confusion matrix for multiple classes diketahui bahwa sistem secara tepat memberikan informasi yang diminta pada smartphone 8 MP dinilai lebih baik, yaitu 90,99% daripada yang ada pada smartphone 5 MP sekitar 88,20%. Sistem berhasil memperoleh informasi kembali pada smartphone 8 MP dengan tingkat recall sebesar 91,01% dinilai lebih unggul, daripada yang dimiliki smartphone 5 MP yang hanya mencapai 87,59%. Tingkat terdekat antara nilai prediksi sistem terhadap nilai aktual lebih baik pada smartphone 8 MP mencapai 91,25%, sementara pada smartphone 5 MP sekitar 88,75%. Kedua smartphone tersebut berada pada tingkat specificity 65% untuk smartphone 8 MP dan 65,21 pada smartphone 5 MP. Berdasarkan evaluasi hasil klasifikasi sistem pada smartphone 8 MP dan 5 MP terhadap hasil klasifikasi secara visual menunjukkan bahwa bahwa tingkat presisi sistem pada smartphone 8 MP dinilai lebih baik, yaitu 88,19% daripada yang ada pada smartphone 5 MP sekitar 84,61%. Tingkat recall sistem pada smartphone 8 MP mencapai 88,25% dinilai lebih unggul, daripada recall pada smartphone 5 MP yang hanya 83,83%. Akurasi sistem pada smartphone 8 MP sekitar 88,75%, sementara pada smartphone 5 MP sebesar 85%. Sistem pada smartphone 8 MP memiliki tingkat specificity mencapai 63,12%, sedangkan pada smartphone 5 MP sebesar 65,09%. Waktu komputasi kinerja sistem yang dihasilkan pada setiap smartphone berbeda-beda tergantung spesifikasi smartphone yang digunakan, yaitu untuk smartphone 1 rata-rata sebesar 10,137 detik, sedangkan pada smartphone 2 sebesar 29,625 detik.
English Abstract
Color analysis of rice leaves is a way to idenfify the nutrient content needed as a basis for recommending fertilizer dosing on rice plants. If there is excess nitrogen, the rice plants are susceptible to disease pests in addition to contaminating ground water. Conversely, if the plant lacks nitrogen, then the growth becomes abnormal. The purpose of this study was to design a system for identifying nitrogen fertilizer dosages based on the greenness of the leaves of rice plants through the concept of image processing using Histogram of s-RGB algorithm and Fuzzy Logic based on android. In this study, Leaf Color Chart (LCC) is a basic concept in the process of developing and designing this system. The system is designed based on 4 scales according to the color level of the LCC in order to identify the image of rice leaves as a basis for recommending the dose of nitrogen fertilizer needed by rice plants. Tests were carried out on two smartphones with different camera resolution capacities, 8 MP and 5 MP smartphones. Based on the test results using the euclidean distance method, it is known that the closest distance of the average RGB value of the system to the value of RGB on LCC is 12,61 on an 8 MP smartphone, while a 5 MP smartphone is 13,97. Confusion matrix for multiple classes evaluation results show that the accuracy of the system to provide the requested information on an 8 MP smartphone is considered better, which is 90,99% compared to what the system has on a 5 MP smartphone of 88,20%. The success of the system to find information back on 8 MP smartphones with a recall rate of 91,01% is considered superior, compared to the success of the system on 5 MP smartphones which only reached 87,59%. The level of closeness between the predictive value of the system and the actual value is better for an 8 MP smartphone of 91,25%, while for a 5 MP smartphone it reaches 88,75%. Both smartphones are at a 65% specificity level for 8 MP smartphones and 65,21 on 5 MP smartphones. Based on the evaluation of the system classification results on 8 MP and 5 MP smartphones on the results of the visual classification shows that the precision level of the system on an 8 MP smartphone is considered better, which is 88,19% compared to those owned by 5 MP smartphones at 84,61%. The rate of recall of the 8 MP smartphone at 88,25% is considered superior, compared to the recall value of the 5 MP smartphone which only reached 83,83%. The system accuracy rate for 8 MP smartphones is 88,75%, while for 5 MP smartphones it reaches 85%. The system on an 8 MP smartphone has a specificity level of 63,12%, while for a 5 MP smartphone it is 65,09%. The computing time of the system performance produced on each smartphone varies depending on the specifications of the smartphone used, for smartphones 1 an average of 10.137 seconds, while for smartphones 2 the average is 29.625 seconds.
| Item Type: | Thesis (Magister) |
|---|---|
| Identification Number: | TES/006.3/SED/i/2019/041903437 |
| Uncontrolled Keywords: | Histogram of s-RGB, Fuzzy Logic, Euclidean Distance, Confusion Matrix for Multiple Classes,Histogram of s-RGB, Fuzzy Logic, Euclidean Distance, Confusion Matrix for Multiple Classes |
| Subjects: | 000 Computer science, information and general works > 006 Special computer methods > 006.3 Artificial intelligence |
| Divisions: | S2/S3 > Magister Teknik Elektro, Fakultas Teknik |
| Depositing User: | soegeng sugeng |
| Date Deposited: | 18 Aug 2022 03:22 |
| Last Modified: | 18 Aug 2022 03:22 |
| URI: | http://repository.ub.ac.id/id/eprint/193294 |
![[thumbnail of Raimundus Sedo.pdf]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text
Raimundus Sedo.pdf Download (6MB) |
Actions (login required)
 |
View Item |