Non-invasive 3D object size measurement system

Date
2014
Authors
Maute, Chris
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract

A point cloud is a 3D representation of the shape of an object. Methods of scanning objects in order to create a point cloud are numerous. We use an effective and inexpensive method called ``structured light'' which utilizes one camera and one projector. The structured light scheme we use is Gray code combined with phase scanning. This scheme utilizes a technique called phase measuring profilometry to reconstruct a high resolution point cloud. Once the point cloud is created, geometric calculations can be performed on the scanned object. We attempt to use point clouds to determine the volume of objects using point clouds that represent only one face of the objects. We achieve this by representing the object as a series of 2D slices and fitting the points of each slice to an circle in order to estimate the area of each slice. We demonstrate the effectiveness of this algorithm by scanning various food items and comparing them to their true volumes. We suggest ways to improve the current algorithm along different approaches to the problem.

Description
This item is available only to currently enrolled UTSA students, faculty or staff.
Keywords
3D vision, computer vision, structured light, volume estimation
Citation
Department
Electrical and Computer Engineering