IDENTIFIKASI MUTU FISIK BIJI PALA (Myristica fragrans houtt) DENGAN TEKNOLOGI PENGOLAHAN CITRA DIGITAL DAN JARINGAN SARAF TIRUAN
Classification process to separate by class quality nutmeg requires great cost and time so that farmers have not done this level, the price of nutmeg low level of farmers. At the level of classification is done by separating the merchants intact seeds and seeds damaged, how it has the disadvantage c...
Saved in:
| Main Authors: | , |
|---|---|
| 格式: | Theses and Dissertations NonPeerReviewed |
| 出版: |
[Yogyakarta] : Universitas Gadjah Mada
2012
|
| 主題: | |
| 在線閱讀: | https://repository.ugm.ac.id/101040/ http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=57056 |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 機構: | Universitas Gadjah Mada |
| 總結: | Classification process to separate by class quality nutmeg requires great cost and time so
that farmers have not done this level, the price of nutmeg low level of farmers. At the level of
classification is done by separating the merchants intact seeds and seeds damaged, how it has the disadvantage can not be carried out continuously, physical fatigue, and high subjectivity. One attempt to improve the classification process is a method of classification is non-destructive, so that the classification process can be done effectively and efficiently. Image processing
technology has been successfully developed for the classification of agricultural products. Image
processing technology can provide good information when combined with the decision-making
system of artificial neural networks.
This research is needed to make the physical quality of the application program
identification nutmeg with image processing and neural networks are able to identify three
classes of nutmeg in the quality of the ABCD, Rimpel and BWP (Broken Wormy Punky). The
method used is image processing, discriminant analysis and artificial neural network using
Matlab programming language. Image processing includes color analysis of RGB, HSV and Lab,
analysis of the form covers the area, perimeter, roundness and compactness, texture analysis
include contrast, correlation, energy, and entropy homogenity. Selection of network input
parameters is done by discriminant analysis.
The results showed the mean parameter S, area, correlation and entropy identified the
quality of nutmeg use feedforward backpropagation network training with the architecture of
four inputs, one hidden layer with 8 neurons, and an output layer. The testing process showed an application program using a GUI (Graphical User Interface) has an accuracy of 100% in the
process of identifying the quality of nutmeg. |
|---|