SISTEM PENGENALAN POLA MOTIF BATIK PADA PERANGKAT ANDROID DENGAN JARINGAN SYARAF TIRUAN; BATIK MOTIF PATTERN RECOGNITION SYSTEM ON ANDROID DEVICE WITH NEURAL NETWORK

A.S., P. SULISTYO, Sri Hartati

2016 | Disertasi | FMIPA

This study aims to identify the image of batik motif with The Learning Vector Quantization (LVQ) methods which takes the best weight training of some experiments by changing the value of the input data, learning rate, number of neurons, and refers to the percentage success rate of the confusion matrix. The Training data that used in this study came from the image motif that has been cropped, and processed with the Canny edge detection, and extracted the features by the edge frequency method. Motif recognition applications show the performance with an 100 % accuracy of the 50 data taken from preliminary data LVQ network training. Further testing using 50 data taken directly from the Android device's camera.The results of the test also has a 100% accuracy. The test results with random data (not pattern motif drilled) indicates accuracy decreased, from 10 data input data is correct only 6 or 60%.

Kata Kunci : batik motif, android devices, neural networks, Learning Vector Quantization (LVQ)