Analisis Pengenalan Pola Daun Berdasarkan Fitur Canny Edge Detection dan Fitur GLCM Menggunakan Metode Klasifikasi k-Nearest Neighbor (kNN)

Leaf Pattern Recognition using Canny Edge Detection and GLCM with k-Nearest Neighbor (kNN)

Abstract

Leaves are one of the parts of plants that can be applied in the process of identifying species images of leaves. The process of classification of leaf imagery can be done by identifying the image of the leaf shape that can be done by identifying the pattern of the leaf by recognizing the structural characteristics of the leaf such as the shape and texture of the leaf. The classification of this leaf image is based on the canny edge detection and gray-level co-occurrence matrix (GLCM) features using the k-Nearest Neighbor (kNN) classification method. The data used is as many as 350 leaf imagery with seven different species. The results show that from the two extraction features used, the Canny feature gets an accuracy of 80% and, the GLCM features gets an accuracy of 93.3%. And the merging of the two features resulted in an increased accuracy of 98%. It can be concluded that this research has produced good accuracy in identifying leaf imagery based on canny edge detection features and Gray-Level Co-occurrence Matrix (GLCM) features and k-Nearest Neighbor classifier method.

Key words -- Leaf, canny edge detection, gray-level co-occurrence matrix (GLCM), the k-nearest neighbor (kNN)