An Improvement Approach Based on the Label Correlation for Automatic Image Annotation

Abstract

Since the semantic gap between low-level visual features and high level-image semantic, the performance of many existing annotation approaches is not satisfactory. In order to bridge the gap and improve the annotation performance, in this paper, an improvement approach based on the measure of label correlation is proposed. According to proposed approach, we can easily measure the correlation between labels. The experimental results confirm that the proposed approach of label set based on the measure of label correlation can improve the efficiency of automatic image annotation systems and achieve better annotation performance than the existing Automatic Image Annotation (AIA) approaches. 