Automatic Image Annotation Based on Sparse Representation and Multiple Label Learning

Abstract

Automatic image annotation has emerged as an important research topic due to its potential application on both image understanding and web image search. Due to the inherent ambiguity of image-label mapping, the annotation task has become a challenge to systematically develop robust annotation models with better performance. In this paper, we present an image annotation framework based on Sparse Representation and Multi-Label Learning (SCMLL), which aims at taking full advantage of Image Sparse representation and multi-label learning mechanism to address the annotation problem. We first treat each image as a sparse linear combination of other images, and then consider the component images as the nearest neighbors of the target image based on a sparse representation computed by L-1 minimization. Based on statistical information gained from the label sets of these neighbors, a multiple label learning algorithm based on a posteriori (MAP) principle is presented to determine the tags for the unlabeled image. The experiments over the well known data set demonstrate that the proposed method is beneficial in the image annotation task and outperforms most existing image annotation algorithms.