Improvement of Automatic Image Annotation Performance Using Bayesian Classifier on Image Retrieval

Abstract

This research proposes Bayesian Classifier to improve image annotation performance on image retrieval. Before being analysed by automatic annotation, descriptions of the image have to be known. Image description is the process of generating descriptions that represent the visual content of images in a certain manner, normally in the form of one or more features. Images are segmented into regions with grid segmentation. Each region are represented by a pre-specifed feature vector. The regions then clustered into a finite set of blobs. The correspondences between the blobs and the words are learned using Statistical Machine Translation and Bayesian Classifier. The experiment result shows that Statistical Machine Translation using Bayesian Classifier can improve precision as compared to Statistical Machine Translation. This method is promising to improve image query result on image retrieval.