Food Calorie Analysis by Using an Automatic Visual Image Mass Measurement Method

• Main Authors: Weng, Wei-Lun, 翁偉倫
• Other Authors: Wen, Bor-Jiunn
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/9c68r9

碩士 === 國立臺灣海洋大學 === 機械與機電工程學系 === 106 === With the technological advances, people pay more attention to health management. In order to develop the automatic food-identification system, this paper uses Bayesian regularization method for identification of food types. This study utilizes a single CCD and infrared depth measurement technology with the Microsoft Kinect sensor to measure of food volume. To establish an automatic visual image mass measurement technology system(AVIMMTS), this study designs a human-machine interface to instantly measure the type, weights, and calories of food by LabVIEW software. AVIMMTS comprises three-step process. Firstly, use a RGB color camera with the Microsoft Kinect sensor to detect the image information of the object. Secondly, the infrared CMOS camera measures the depth information of the object. In addition, the color information of image is utilized to identify the types of the food by using neural network learning of Bayesian regularization method. Thirdly, calculate the weights and the calories of the food by the type and depth information of the object. Eventually, the measurement error of the calorie are within 15% in this study.