Identification of Medicine Leaf Images Using Invariant Moment and K-Nearst Neighboor

Fredianto
Enny Sela (UTY)
Suhirman
Ardhi Wicaksono Santoso

International Journal of Engineering, Technology and Natural Sciences (IJETS)
Vol 3 No 1 (2021): International Journal of Engineering, Technology and Natural Sciences

Medicinal plants have benefits for preventing or curing various diseases. The number of medicinal plants and the lack of knowledge about the types of medicinal plants make it difficult for people to distinguish the types of medicinal plants. This difficultness causes people to prefer to use chemical drugs rather than medicinal plants. This study develops a system of identification of medicinal plants. There are four steps to build the system: input leaf images, pre-processing, invariant moment feature extraction, and K-Nearest Neighbours (K-NN) pattern recognition. A 100 images samples images from 5 types of medicinal plants were involved in this study. The identification process of leaf image begins with the cropping, resizing process, and several morphological operations. Then feature extraction stage uses invariant moments. The final stage of pattern recognition uses K-NN. The result of this research is that the system can identify the types of medicinal plants. Using the Manhattan distance, the study archives the highest average accuracy.