Automatic Image Annotation by Incorporating Weighting Strategy and CSOM Classifier

Automatic image annotation (AIA) emerges in recent years, and it attempts to replace a huge amount of manual efforts for image annotation. In this study, we propose a novel framework incorporating weighting strategy with concurrent self-organizing map (CSOM) classifier based on the concept of classification. Further, we apply this model to determine a classified weight that an image belongs to some specific class and assign appropriate keywords through the mapping table associated with the specific class under weighting strategy. Result shows that the match rate is between 33.33% through 100% and the average correct rate of annotation achieves 78.44% if the image is assigned to the correct class.