Version 1
: Received: 23 June 2018 / Approved: 24 June 2018 / Online: 24 June 2018 (11:19:41 CEST)
How to cite:
Kim, S.; Lee, W. A Real-Time Calibration Method of Kinect Recognition Range Extension for Media Art. Preprints2018, 2018060367. https://doi.org/10.20944/preprints201806.0367.v1
Kim, S.; Lee, W. A Real-Time Calibration Method of Kinect Recognition Range Extension for Media Art. Preprints 2018, 2018060367. https://doi.org/10.20944/preprints201806.0367.v1
Kim, S.; Lee, W. A Real-Time Calibration Method of Kinect Recognition Range Extension for Media Art. Preprints2018, 2018060367. https://doi.org/10.20944/preprints201806.0367.v1
APA Style
Kim, S., & Lee, W. (2018). A Real-Time Calibration Method of Kinect Recognition Range Extension for Media Art. Preprints. https://doi.org/10.20944/preprints201806.0367.v1
Chicago/Turabian Style
Kim, S. and Won-hyung Lee. 2018 "A Real-Time Calibration Method of Kinect Recognition Range Extension for Media Art" Preprints. https://doi.org/10.20944/preprints201806.0367.v1
Abstract
Kinect is a device that has been widely used in many areas since it was released in 2010. Kinect SDK was announced in 2011 and used in many other areas than its original purpose, which was a controller for gaming. In particular, it has been used by a number of artists in digital media art since it is inexpensive and has a fast recognition rate. However, there is a problem. Kinect create 3D coordinates with a single 2D RGB image for x, y value - single depth image for z value. And this creates a significant limitation on the installation for interactivity of media art. Because the Cartesian XY coordinate and the spherical Z coordinate system are used in combination, depth error depending on the distance is generated, which makes real-time rotation recognition and coordinate correction difficult above coordinate system. This paper proposes a real-time calibration method of Kinect recognition range expansion for useful application in the digital media art area. The proposed method can recognize the viewer accurately by calibrating a coordinate in any direction in front of the viewer. 3,400 datasets witch acquire from experiment were measured as five stances: the 1m attention stance, 1m hands-up stance, 2m attention stance, 2m hands-up stance, and 2m hands-half-up stance, which were taken and recorded every 0.5 sec. The experimental results showed that the accuracy rate was improved about 11.5% compared with front measurement data according to Kinect reference installation method.
Keywords
kinect; depth calibration; RGB-D; media art; skeletal joint data
Subject
Computer Science and Mathematics, Hardware and Architecture
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
