PERBANDINGAN TINGKAT PENGENALAN CITRA DIABETIC RETINOPATHY PADA KOMBINASI PRINCIPLE COMPONENT DARI 4 CIRI BERBASIS METODE SVM (SUPPORT VECTOR MACHINE)

Sari Ayu Wulandari (Universitas Dian Nuswantoro)
Rudy Tjahyono (Universitas Dian Nuswantoro)
Dian Retno Sawitri (Universitas Dian Nuswantoro)

Majalah Ilmiah Teknologi Elektro
Vol 15 No 1 (2016): (January - June) Majalah Ilmiah Teknologi Elektro

Abstract— Pattern recoqnition methods for image of diabetic retinopaty are influenced by differences in pigmentation. To help diabetic retinopathy image recognition is required a software. This paper presents the results of research on pattern recognition image of diabetic retinopathy,This study used the image of the yellow canal with Gabor filter.Characteristics that are taken from each image is characteristic of the mean,variance, skewness and entropy, followed by feature extraction with PCA (Principle Component Analysis).At PCA feature extraction, square matrix whose number of columns equal to the number of features is enerated.There are four features used. These features are 4 PCs (Principle Component), ie, PC1, PC2, PC3 and PC4.From the combination of these features, we obtained six pairs that consist of two traits. By using a linear model of SVM will been selected the pair with the highest accuracy value. Based on the analysis, we obtained a couple PC1and PC2 models that have the highest levels of learning (100%) and the fastest recognition time, which is explicitly indicated by the smallest amount of support vector.