Open AccessCONTENT BASED IMAGE RETRIEVAL DENGAN METODE COLOR MOMENT DAN K-MEANS
Author(s) -
I Made Sukafona,
Emmy Febriani Thalib
Publication year - 2018
Publication title -
jurnal resistor (rekayasa sistem komputer)
Language(s) - English
Resource type - Journals
eISSN - 2598-9650
pISSN - 2598-7542
DOI - 10.31598/jurnalresistor.v1i2.322
Subject(s) - artificial intelligence , computer science , content based image retrieval , image retrieval , image segmentation , precision and recall , segmentation , k means clustering , feature (linguistics) , image texture , computer vision , feature extraction , process (computing) , pattern recognition (psychology) , image (mathematics) , cluster analysis , linguistics , philosophy , operating system
Content Based Image Retrieval (CBIR) is a research cluster that is very important to overcome problems related to the image search process. The development of internet technology and data communication has caused the number of multimedia images currently circulating to be very high. This study took the Color Moment method to carry out the feature extraction process. Before the feature extraction process, a segmentation process was carried out to separate the background image and the foreground image. Next, each background and front image is stored in the database. Method performance measurement is done by calculating the value of precision and recall. The test image used is the Wang dataset consisting of ten image classes. The test results show the level of recall or completeness of the images that were found to have increased significantly after using the K-Means segmentation process. But a high enough recall value decreases the value of precision or the comparison of true images with the image found overall. Precision values decrease when compared to the CBIR method without running the K-Means segmentation.