Multi-view model contour matching based food volume estimation

Abstract

In this paper, an automatic food volume estimation method based on outer contour matching is proposed, which avoids the complicated calculation. We pre-defined a simple 3D model library and stored the projections and the user’s hand. Users took three images containing their hands from three views. The contour of segmented image was compared with the projections to find the best match. Meanwhile, we took the user’s hand as the scale and calculated the volume. As the method is easy to operate, less space-consuming, it is quite suitable for integrated application in the mobile app.