A New Texture Feature for Improved Food Recognition Accuracy in a Mobile Phone Based Dietary Assessment System

Abstract

Poor diet is one of the key determinants of an individual's risk of developing chronic diseases. Assessing what people eat is fundamental to establishing the link between diet and disease. Food records are considered the best approach for assessing energy intake however paper-based food recording is cumbersome and often inaccurate. Researchers have begun to explore how mobile devices can be used to reduce the burden of recording nutritional intake. The integrated camera in a mobile phone can be used for capturing images of food consumed. These images are then processed to automatically identify the food items for record keeping purposes. In such systems, the accurate classification of food items in these images is vital to the success of such a system. In this paper we will present a new method for generating texture features from food images and demonstrate that this new feature provides greater food classification accuracy for a mobile phone based dietary assessment system.