Abstract: Image based food identification is an emerging research topic for much industrial application. It refers to the capability of identifying various food items based on the visual information. Unfortunately, food items classification is highly sensitive to the accuracy of the image segmentation which is not always satisfying due to many factors. In this study, an aggregated features association classifier is proposed to handle the resultant problem of non-accurate image segmentation. It uses ELM for food items classification. Also, it exploits the fact that food items are associated with others when they are placed in the plate; the accuracy of the classifier has been improved using features association. An accuracy of 100% is obtained for input images with over or under segmentation errors which proves the usefulness of this algorithm.