Key Features and Differences in the Kinect V2 Support

Differences in the Kinect V2 Support

The Kinect Adaptor lets you acquire images using a Kinect® for Windows® device. Kinect devices are often used in automotive IVS, robotics, human-computer interaction (HCI), security systems, entertainment systems, game design, and civil engineering. Uses of Kinect devices include body analysis, 3-D mapping, gesture recognition, human travel patterns, and sports and games.

Doing image acquisition with a Kinect for Windows camera is similar to using other cameras and adaptors, with these key differences:

  • The Kinect for Windows device has two separate physical sensors, and each one uses a different DeviceID in the videoinput object. The Kinect color sensor returns color image data. The Kinect depth sensor returns depth data and skeletal data can be obtained as frame metadata. For information about Kinect device discovery and the use of two device IDs, see Detect the Kinect V2 Devices.

  • The Kinect for Windows V2 device returns five data streams. The image stream is returned by the color sensor and contains color data in the RGB_1920x1080 color format. The depth stream is returned by the depth sensor and returns depth information in pixels in the DEPTH_512X424 format. The body stream is returned by the depth sensor and returns metadata about the bodies. The audio stream is unused by Image Acquisition Toolbox™. For details on the streams, see Data Streams Returned by the Kinect V2 Adaptor in MATLAB.

  • The Kinect for Windows V2 device can track up to six people with 3-D position and orientation. It also supports hand tracking and state on two bodies.

  • In Image Acquisition Toolbox, you access body metadata through the depth sensor object. For an example showing how to acquire color images and body metadata, see Acquire Image and Body Data Using Kinect V2.

Supported Hardware for the Kinect V2 Support

The Kinect V2 support includes the following supported hardware:

  • Kinect for Windows V2 Sensor

  • Kinect for Windows Developer Bundle, which includes the Kinect Sensor for Xbox One and the Kinect Adapter for Windows

Note

The Kinect adaptor is intended for use only with the Kinect for Windows sensor.

Requirements for the Kinect V2 Support

The following requirements apply to the Kinect V2 support in Image Acquisition Toolbox.

  • The Image Acquisition Toolbox Support Package for Kinect for Windows Sensor must be installed.

  • The Kinect adaptor is supported on 64-bit Windows.

  • The Kinect V2 support requires Windows 8.0.

  • You must use a USB 3.0 host controller to connect your Kinect V2 device to your computer.

  • To use Kinect for Windows V2 support, you must have version 2.0 of the Kinect for Windows Runtime installed on your system. If you do not already have it installed, it installs with the Kinect support package.

  • You can only use one Kinect V2 device at a time with the Image Acquisition Toolbox. You can use a Kinect V2 and a Kinect V1 at the same time, but only one Kinect V2 device at a time. This is a hardware limitation.

Note

With previous versions of the Image Acquisition Toolbox, the files for all of the adaptors were included in your installation. Starting with version R2014a, each adaptor is available separately through support packages via MATLAB® Add-Ons. In order to use the Image Acquisition Toolbox, you must install the adaptor that your camera uses, in this case, the Image Acquisition Toolbox Support Package for Kinect for Windows Sensor.

Features of the Kinect V2 Support

Kinect for Windows V2 offers these new features:

  • Higher resolution capabilities, including three times more depth fidelity and a cleaner depth map

  • Wider depth and color field of view

  • A 25-point skeleton for each of up to six people (Kinect V1 has 20 joints)

  • Tracks up to six people simultaneously (Kinect V1 tracks only 2)

  • Open-hand and closed-hand gesture recognition

  • Biocorrect body joints (particularly hip, shoulder, and spine)

  • Higher confidence for joints and more points for hands