Classification of Zingiber Plants Based on Stomate Microscopic Images Using Probabilistic Neural Network (PNN) Algorithm

Abstract Herbal plants are medicinal plants included in the first choice for medicinal ingredients for rural residents. Herbal plants such as ginger and lime are plants that are often used and have plants that are similar in level. One way to identify these types of plants is through stomata microscopic imagery. As for the inspection that is still done manually by the pharmacists, the manual inspection takes a long time and misidentification may occur because there are some similarities with direct viewing, so a method is needed to identify the types of herbal plants based on stomata microscopic images automatically and to improve accuracy in the identification process. The method proposed in this study is Probabilistic Neural Network for the identification of herbal plant types based on stomata microscopic images and the algorithm used to identify these plant types Probabilistic Neural Network (PNN). Before the identification stage, the image will go through three stages, namely preprocessing, segmentation and feature extraction using the Gray level co-occuration matrix method. After testing using 20 microscopic image data of ginger and bitter ginger stomata. It was concluded that the proposed method has the ability to identify types of herbal plants with an accuracy percentage of 90 %.