Javanese Letters Recognition Using Canny Edge Detection, Principal Component Analysis, and Support Vector Machine (SVM)

The Javanese letters are one of the cultural heritage of the Javanese tribe with unique letter shapes and writing procedures. This uniqueness makes the Javanese letters an interesting material in pattern recognition research. Canny edge detection, Principal component analysis (PCA), and Support vector machine (SVM) are methods commonly used and considered to be superior in pattern recognition. This research was conducted to determine the ability of Canny edge detection, PCA, and SVM in recognizing Javanese letters. Canny edge detection and PCA are used to perform feature extraction. While SVM will be used as a method of recognizing Javanese letters in this study. Through this research, the performance of Canny edge detection, PCA, and SVM methods was obtained in recognizing 20 Javanese letters. The best average accuracy of recognizing Javanese letters is 91.582% using Canny, PCA, and SVM. PCA and SVM have 91.500% accuracy. The lowest average accuracy is 89.917% using Canny and SVM.