Perbandingan Classifier untuk Identifikasi Citra Tanaman Hias
Abstract
The use of an appropriate classification technique significantly determines the result of the identification process. In this research, we perform comparison of classification techniques for the identification of ornamental plants. The classification techniques used are Probabilistic Neural Network (PNN), K-Nearest Neighbor (KNN), Naive Bayes (NB) and the combination of K- Nearest Neighbor and Naive Bayes. We use 300 images of ornamental plants, consisting of 30 different types of plants. Plant image features are extracted using texture extraction technique, called Local Binary Pattern (LBP). The best performance was shown by the classification technique using PNN, KNN, and the combination of KNN and NB, with an accuracy of 72.22%. This result provides a promising prospect for an effective and efficient ornamental plants identification system.
Collections
- UT - Computer Science [2236]