Penggabungan Ciri Citra dan Teks untuk Sistem Pencarian Tumbuhan Obat Indonesia Berbasis Web

Abstract

This research investigate the effectiveness of image and text fusion for Indonesian medicinal plant search engine based on web application. This research used 51 species of Indonesian medicinal plants and each species consists of 48 images, so the total images used in this research are 2448 images. WeightedSUM is general linear combination formula to combine image and text features. Further research was conducted on the combination of features to get a better result in medicinal plants search engine. Fuzzy Local Binary Pattern (FLBP) is used to extract texture in image processing. This research uses the Probabilistic Neural Network (PNN) to improve image similarity that used in the fusion process. BM25 weighting in document search engine is used to get text similarity. Image and text similarity are combined using WeightedSUM to get the retrieval results. The experimental results show that the fusion of image and text features can improve the performance of retrieval results. In particular, the Average Precision (AVP) has increased from 0.31 to 0.71