Volume Approximation Using Kinect Sensor

Abstract

Kinect has already contributed to object detection, 3D modeling, autonomous navigation, and scene mapping studies. This research aims to use Kinect’s ability to collect depth data by approximating the volume capacity of an open-top subject and producing a 3D representation of it based on the data collected from Kinect and image processing. The experiment included taking the subject’s ground truth volume and comparing it to the system’s volume output. Using Linear Regression for the data interpretation indicates that the device created is reliable enough to produce a correlation coefficient of 0.9621. A significant positive association of the two datasets: experimental volume and theoretical volume. The prototype arrived with an average error rate of 9.003%, implying that the system can get accurate results.