Dietary Intake Estimation by Using Colored Point Cloud Processing

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K20267
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 90150:Medical assistive technology-related
• Research Institution: Aichi Prefectural University
• Principal Investigator: Suzuki Takuo 愛知県立大学, 情報科学部, 准教授 (80709303)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 点群処理 / 形状近似 / 欠損補完 / 体積推定 / 食事管理 / 生活支援

Outline of Final Research Achievements

In this research, the principal investigator proposed a point cloud processing method for accurately estimating dietary intake with an RGB-D camera. In the case of a convex object such as a spherical object, a dropout of points occurs at the outer edge, and the estimated volume will be a bit small.
In the proposed method, a plane-symmetrical point cloud is generated for the acquired point cloud and synthesized with the acquired point cloud. Then, the synthesized point cloud is approximated by implicit functions. If there is a dropout around the outer edge, the approximated shape will be the sphere with a hole, so the presence or absence of a dropout can be determined based on the presence or absence of the hole. Also, if the range of a dropout is small, the hole can be closed when approximating with implicit functions, which means that the missing part can be complemented with points.

Free Research Field

知能ロボティクス

Academic Significance and Societal Importance of the Research Achievements

本研究では面対称な点群を用いることで外挿でなく内挿で点群を補完できることを示した。これは安定して高精度に欠損部を補完できることを意味し、食物体積推定、更にはカロリーや栄養素の摂取量推定の高精度化に貢献したと言える。
提案手法では物体の外縁部に点群を発生させたが、その延長線上で物体の裏側に点群を発生させることができるかもしれない。RGB-Dカメラから見える側の形状に基づいて見えない側の形状を推定する技術の一助になればと考えている。