Food Image Retrieval with Gray Level Co-Occurrence Matrix Texture Feature and CIE L*a*b* Color Moments Feature

Food recognition and finding its recipe sometime is becoming challenging in everyday life, especially those who work in cuisine field. Most people usually try to find the food recipe using search engine or specially-designed recipe website by typing the query in the form of text. The problem is, text query cannot be used to accommodate the need of the user when they try to find the food and its recipe using the food image. Hence, a specialized content-based image retrieval is needed for food finding. This paper proposes food image retrieval that later can be matched with its associated recipe. The features used in this food image retrieval is texture and color moments features. Texture feature used in this research is Gray-Level Co-occurrence Matrix (GLCM) while CIE L*a*b color moment is used in color feature extraction. From the initial ranked result, using small dataset consists of 1303 training data and 31 test data, the proposed system can achieve good result up to 97.6% of Mean Average Precision in top-10 rank. We are also investigating four different proximity measures, namely Euclidean, Manhattan, Minkowski, Canberra distance and resulted in Minkowski as the best proximity measure in this dataset.