Hybrid of Wavelet Feature Extraction and LVQ Neural Network to Recognize Patchouli Variety using Leaf Images

Candra Dewi

2019

Abstract

Patchouli consist of some varieties that have different patchouli alcohol (PA). This variety can be recognized by experts who dabbling with patchouli plants through observation of shape and texture of the leaf. This study introduced a new method to identify patchouli varieties by utilizing leaf images. The wavelet feature extraction was used to obtain leaf texture characteristics. The varieties then are identified by using Learning Vector Quantization (LVQ) Neural Network algorithm. The results of testing on 40 leaf image data showed the value of recognition accuracy of patchouli varieties reached 83, 33%. This result is obtained by wavelet parameters namely doubechies level 3, doubechies coefficient 3, and LVQ parameters, namely learning rate 0.1 learning rate reduction constant 0.2. These results can be said to be quite good considering that the patchouli leaf tested have almost similar shape and color.

Paper Citation

in Harvard Style

Dewi C. (2019). Hybrid of Wavelet Feature Extraction and LVQ Neural Network to Recognize Patchouli Variety using Leaf Images.In Proceedings of the 2nd International Conference of Essential Oils - Volume 1: ICEO, ISBN 978-989-758-456-5, pages 22-28. DOI: 10.5220/0009954800220028

in Bibtex Style

@conference{iceo19,
author={Candra Dewi},
title={Hybrid of Wavelet Feature Extraction and LVQ Neural Network to Recognize Patchouli Variety using Leaf Images},
booktitle={Proceedings of the 2nd International Conference of Essential Oils - Volume 1: ICEO,},
year={2019},
pages={22-28},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0009954800220028},
isbn={978-989-758-456-5},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference of Essential Oils - Volume 1: ICEO,
TI - Hybrid of Wavelet Feature Extraction and LVQ Neural Network to Recognize Patchouli Variety using Leaf Images
SN - 978-989-758-456-5
AU - Dewi C.
PY - 2019
SP - 22
EP - 28
DO - 10.5220/0009954800220028