Optimasi Metode K-Nearest Neighbor Dengan Particle Swarm Optimization Untuk Pengenalan Citra Batik Dengan Ragam Hias Geometris

• Published in: Jurnal informatika UPGRIS (Online) Vol. 8; no. 1
• Main Authors: Widyatmoko, Karis, Sugiarto, Edi, Muslih, Muslih, Budiman, Fikri
• Format: Journal Article
• Language: English; Indonesian
• Published: Universitas PGRI Semarang 19-07-2022
• Subjects: batik, particle swarm optimization, k-nearest neighbor
• ISSN: 2460-4801; 2477-6645
• DOI: 10.26877/jiu.v8i1.11705

Batik is an intangible cultural heritage that has been approved by UNESCO as an Indonesian cultural heritage. Batik is the art of drawing on cloth for clothing. This art of drawing is in the form of a pattern and has a philosophical meaning which is historically very closely related to the philosophy of Javanese culture. The various patterns of this batik need to be preserved, and one of the efforts to preserve this diversity is to maintain the special characteristics of the batik motif and continue to introduce it to future generations. Efforts to facilitate the introduction of batik motifs can be done through technology that is able to automatically recognize batik patterns according to the motif being tested, one of which is pattern recognition technology. This study aims to optimize the KNN method with PSO to increase the accuracy of batik pattern recognition, the research was carried out in several stages starting from data collection, preprocessing, feature extraction, and classification. The research was conducted using 310 data in the form of batik images in 7 different motifs and divided into 240 compositions for training data and 70 for testing data. At the feature extraction stage, the discrete wavelet transform method is used up to level 3, then at the classification stage, the PSO and KNN algorithms are used. The PSO algorithm is used to obtain the most optimal number of k which is used as the input parameter k in the KNN algorithm. The results of the research that have been carried out have proven that with the addition of the PSO algorithm, the accuracy of the KNN method can be increased by 6% compared to the standard KNN method.