Automatic Meal Inspection System Using LBP-HF Feature for Central Kitchen

This paper proposes an intelligent and automatic meal inspection system which can be applied to the meal inspection for the application of central kitchen automation. The diet specifically designed for the patients are required with providing personalized diet such as low sodium intake or some necessary food. Hence, the proposed system can benefit the inspection process that is often performed manually. In the proposed system, firstly, the meal box can be detected and located automatically with the vision-based method and then all the food ingredients can be identified by using the color and LBP-HF texture features. Secondly, the quantity for each of food ingredient is estimated by using the image depth information. The experimental results show that the meal inspection accuracy can approach 80%, meal inspection efficiency can reach1200ms, and the food quantity accuracy is about 90%. The proposed system is expected to increase the capacity of meal supply over 50% and be helpful to the dietician in the hospital for saving the time in the diet inspection process.