Open AccessAlgorithm Configuration K-Nearest To Clarification Medicine Tree Based On Extraction, Variation Of Color, Texture And Shape Of Leaf
Author(s) -
Ardhi Dinullah Baihaqie,
Rayung Wulan
Publication year - 2021
Publication title -
ilomata international journal of social science
Language(s) - English
Resource type - Journals
eISSN - 2714-8998
pISSN - 2714-898X
DOI - 10.52728/ijss.v2i1.187
Subject(s) - hue , mathematics , rgb color model , pattern recognition (psychology) , artificial intelligence , texture (cosmology) , horticulture , botany , computer science , image (mathematics) , biology
At this time to overcome difficulties in identifying medicinal plants that have an impact on the frequent errors in the use of medicinal plants. The formulation of the problem to be discussed in this study is how to identify medicinal plants based on feature extraction of color, texture, and leaf shape. Steps to resolve this problem by collecting image data of medicinal plants, then the image data extracted leaf color features using Red Green Blue (RGB) and Hue Saturation Value (HSV), based on leaf texture using the Gray Level Co-occurrence Matrix (GLCM), based on the shape leaves use eccentricity and metrics. and then classified by the K-Nearest Neighbor (KNN) method. The results in this study the accuracy of Chinese Petai leaves is superior to other types of leaves, which is 98%, which occurs at each K value. Other types of leaves have various values. Saga leaves range between 94% - 97%, Green Betel leaves between 92.8% - 97%, and Red Betel leaves between 91.7% - 95%, Optimal K values indicated by K = 3 have an average accuracy rate of 96.7% also have sensitivity value of 93.3%. The addition of K = 5, K = 7, K = 9, and K = 11 tends to decrease the average value of accuracy and sensitivity.