Color Based Identification and Classification of Boiled Food Grain Images

Texture and color are the important features used in identifying objects or regions of interest (ROI) in any image, be it a photomicrograph, an aerial photograph, or a satellite image. We propose a methodology for identification and classification of boiled food grains based on the level of boiling using two color models HSV and L*a*b*, in Indian context. These color models provide good texture definition for any image. The classification is performed at two levels: Level 1 determines the type of grain image and Level 2 estimates the amount of boiling, as full boiled, medium boiled and half boiled. Results show average accuracies of 80% and 96% for first level and 70% and 96% accuracies for second level classifications for HSV and L*a*b*, respectively. The work is applicable to automatic inspection of food preparations in food industry and monitoring, cooking, and serving food in restaurants, hotels and malls by service robots.