Implementasi Algoritma Freeman Chain Code dan Algoritma K-Nearest Neighbor Dalam Pengenalan Huruf Mandarin

Abstract

Mandarin is the second international language used by the world's population and is the language most studied by students in Indonesia. This research was conducted to determine the success rate of the Freeman Chain Code algorithm and the K-Nearest Neighbor algorithm in mandarin letter recognition. In recognizing Chinese characters, there are several stages that must be passed, namely pre-processing, feature extraction, and letter recognition. The pre-processing stage uses grayscale, Gaussian Blur, binaryization, and thinning. The feature extraction stage uses the Freeman Chain Code Algorithm and the Depth First Search (DFS) Algorithm. The classification stage uses the K-Nearest Neighbor Algorithm and the L1-Metric Algorithm (Manhattan Distance). In this study, there were 10 letter classes with each letter having 100 sample images. The distribution ratio of this research is 70% training data and 30% testing data. This research produces an application that is able to recognize Chinese characters. The success rate resulting from this study was 72% with 216 of the 300 images successfully recognized