Dahlena, Elda (2018) Prediksi Klorofil Sayuran Dengan Metode Jaringan Saraf Tiruan Berbasis Citra Digital. Sarjana thesis, Universitas Brawijaya.
Abstract
Sayuran berklorofil sangat penting bagi kesehatan, karena mengandung sumber nutrisi. Kualifikasi visual yang kurang objektif, serta analisis laboratorium yang lama dan cukup mahal untuk setiap sampelnya, mendorong solusi alternatif dari metode jaringan saraf tiruan berbasis citra digital untuk diterapkan dalam memprediksi klorofil total pada sayuran daun. Hasil penelitian menunjukkan bahwa Red, saturation_HSV, lightness dan saturation_HSL terhadap klorofil total, a dan b menghasilkan liniearitas positif. Green, blue, hue,dan value menghasilkan linieritas negatif. Regresi positif tertinggi ditunjukkan pada indeks saturation_HSL terhadap klorofil total daun singkong sebesar 78,6 %. Regresi negatif tertinggi terjadi pada indeks green terhadap klorofil total daun kangkung sebesar 49,8 %. Model terbaik yang dihasilkan melalui metode jaringan saraf tiruan dalam memprediksi klorofil total adalah model jaringan dengan 8 input, 9 hidden layer dan satu output layer, pada proporsi data training 75 % dan data testing 25 % sehingga diperoleh nilai MSE testing terkecil sebesar 0,0921, dengan Regresi testing sebesar 0,8468. Model jaringan mampu membaca sumber klorofil tertinggi pada daun singkong dengan presentase 84,68 %.
English Abstract
Chlorophyll’s vegetable is very important for health, because it contain a source of nutrients. Less objective visual qualifications, laboratory analyzes need long time and expensive for each sample, encourage alternative solutions of image processing based with neural network methods to be applied at prediction total chlorophyll in leaf’s vegetable. The results showed that Red, saturation_HSV, lightness and saturation_HSL to total chlorophyll, a and b yielded positive linearity. Green, blue, hue, and value yielded negative linearity. The highest positive regression was 78.6%, shown on the saturation_HSL of total chlorophyll cassava leaves . The highest negative regression occurred on green mean to total chlorophyll of leaf kangkung by 49.8%. The best model artificial neural network in predicting total chlorophyll is network model with 8 inputs, 9 hidden layers and one output layer, with 75% training data and 25% testing data proportion, so as to obtain the smallest MSE testing is 0.0921, with regression testing 0.8468. This Network model is able to read the highest source of chlorophyll on cassava leaves with percentage of 84.68%.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FTP/2018/270/051807906 |
| Uncontrolled Keywords: | Citra Digital, Klorofil, MSE, Regresi, Sayuran / Chlorophyll, Image Processing, MSE, Regretion, Vegetable |
| Subjects: | 300 Social sciences > 338 Production > 338.1 Agriculture > 338.16 Production efficiency > 338.162 Agricultural methods |
| Divisions: | Fakultas Teknologi Pertanian > Keteknikan Pertanian |
| Depositing User: | Endang Susworini |
| Date Deposited: | 31 Oct 2019 02:27 |
| Last Modified: | 30 Jun 2022 08:07 |
| URI: | http://repository.ub.ac.id/id/eprint/165921 |
![[thumbnail of Elda Dahlena.pdf]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text
Elda Dahlena.pdf Download (4MB) |
Actions (login required)
 |
View Item |