202 健康管理ロボットによる食事摂取量計測に関する基礎的研究(ライフサポートにおける工学技術)

概要

論文の詳細を見る
Healthcare in daily life plays a significant role in preventing lifestyle-related diseases. We have been developing a health care robot which measures vital signs, meal consumption, physical activity to manage human's health. A three dimensional (3D) reconstruction method based on multiple images treatment with a reference object is proposed to measure the meal volume as consumption. The robot records multiple images of the foods and measures the meal volume by using the 3D reconstruction method. Following results are obtained in this study. Firstly, the increasing number of treatment images, from five images to twelve images, reduces the reconstruction error from 5 mm to 4 mm. The multiple images treatment is good to reduce the reconstruction error. Second, the reconstruction error due to the reference object's angle widely ranges from 4 mm to 45 mm. The image angle of the reference object affects the reconstruction accuracy significantly.
社団法人日本機械学会の論文
2007-09-28