Klasifikasi Citra dengan Support Vector Machine pada Sistem Temu Kembali Citra
Abstract
This paper presents a Support Vector Machine (SVM) algorithm for image classification in Content Based Image Retrieval (CBIR). Expectation Maximization is applied to image segmentation. The classification is based on color image. The algorithm is optimized using Sequential Minimal Optimization (SMO). This research is also compared to the Euclidean distance for measurement image similarity. The experiment on Caltech Database and images from www.flower.vg shows that the algorithm can improve the performance. The average precision of the experiment using SVM and SMO is 76,76% while Euclidean distance is 50.91%. This algorithm is a promising algorithm to be used for Content Based Image Retrieval (CBIR).