Point Cloud Processing Method for Food Volume Estimation Based on Dish Space

It is necessary to know the amount of food on dishes in order to encourage taking medicine after eating. Also, for health management, it is vital to record what and how much a person ate. Although there are research cases using weight sensors or color cameras, it has been difficult to estimate the food volume accurately and inexpensively at home. In previous works, the authors developed a technique for estimating volume based on a depth image acquired by a depth camera. In this paper, the authors propose a new point cloud processing method for a more accurate estimation. A point cloud is a set of coordinate points on objects and is suitable for processing objects three-dimensionally. The authors have developed a technique for recognizing dishes on the dining table based on a point cloud and constructing the dish space. Additionally, another technique was developed for estimating the volume of food in the dish space.