Different ways of retrieving the color stream from Kinect
by Abhijit Jana
Kinect for Windows SDK Programming Guide
- Kinect for Windows SDK Programming Guide
- Table of Contents
- Kinect for Windows SDK Programming Guide
- Credits
- About the Author
- Acknowledgement
- About the Reviewers
- www.PacktPub.com
- Preface
- 1. Understanding the Kinect Device
- 2. Getting Started
- System requirements for the Kinect for Windows SDK
- Evaluation of the Kinect for Windows SDK
- Downloading the SDK and the Developer Toolkit
- Installing Kinect for Windows SDK
- Testing your device
- Looking inside the Kinect SDK
- Features of the Kinect for Windows SDK
- Capturing the color image data stream
- Processing the depth image data stream
- Capturing the infrared stream
- Tracking human skeleton and joint movements
- Capturing the audio stream
- Speech recognition
- Human gesture recognition
- Tilting the Kinect sensor
- Getting data from the accelerometer of the sensor
- Controlling the infrared emitter
- The Kinect for Windows Developer Toolkit
- Making your development setup ready
- The Coding4Fun Kinect Toolkit
- Summary
- 3. Starting to Build Kinect Applications
- How applications interact with the Kinect sensor
- Kinect Info Box – your first Kinect application
- Dealing with the Kinect status
- Summary
- 4. Getting the Most out of Kinect Camera
- Understanding the Kinect image stream
- Different ways of retrieving the color stream from Kinect
- KinectCam – a Kinect camera application
- Enabling the color stream channel
- Looking inside color image stream helpers
- Capturing frames on demand
- Extending the KinectCam
- Applying more effects to the camera
- Seeing in low light
- Making your application perform better
- Using the Coding4Fun toolkit
- Summary
- 5. The Depth Data – Making Things Happen
- Understanding the depth data stream
- Capturing and processing depth data
- Looking inside depth image stream helpers
- Depth data and distance
- Working with depth range
- Depth data distribution
- Player index with depth data
- Getting the depth and player index automatically
- A 3D view of depth data
- Summary
- 6. Human Skeleton Tracking
- How skeleton tracking works
- Skeleton tracking with the Kinect SDK
- Start tracking skeleton joints
- Flow – capturing skeleton data
- An intrusion detector camera application
- Looking inside skeleton stream helpers
- Skeleton-tracking mode
- Skeleton tracking in near mode
- The Skeleton
- Choosing which skeleton to track
- The building blocks – Joints and JointCollection
- Steps to be followed for joint tracking
- Create your own joints data point
- Bones – connecting joints
- Adjusting the Kinect sensor automatically and giving live feedback to users
- Skeleton smoothing – soften the skeleton's movement
- Skeleton space transformation
- The Advanced Skeleton Viewer application
- Debugging the applications
- Getting data frames together
- Summary
- 7. Using Kinect's Microphone Array
- Verifying the Kinect audio configuration
- Using the Kinect microphone array with your computer
- The Kinect SDK architecture for Audio
- Kinect microphone array
- Audio signal processing in Kinect
- Taking control over the microphone array
- Kinect sound recorder – capturing Kinect audio data
- Processing the audio data
- Sound source localization
- Summary
- 8. Speech Recognition
- 9. Building Gesture-controlled Applications
- What is a gesture
- Approaches for gesture recognition
- Basic gesture recognition
- Algorithmic gesture recognition
- Which gestures can be considered as algorithmic
- Understanding the algorithmic gesture detection approach
- Implementing an algorithmic gesture
- Adding gesture types
- Extending the Event argument
- Adding a GestureHelper class
- Defining the GestureBase class
- Implementing the SwipeToLeftGesture class
- Adding the ZoomIn, ZoomOut, and SwipeToRight gesture classes
- Implementing the GestureRecognitionEngine class
- Using the GestureRecognitionEngine class
- A demo application
- Making it more flexible
- Weighted network gesture recognition
- Template-based gesture recognition
- Building gesture-enabled controls
- The Basic Interaction – a WPF application
- Key things to remember
- Summary
- 10. Developing Applications Using Multiple Kinects
- Setting up the environment for multiple Kinects
- Multiple Kinects – how to reduce interference
- Detecting multiple Kinects
- Developing an application with multiple Kinects
- Controlling multiple sensor status changes
- Handling a failover scenario using Kinects
- Challenges faced in developing applications using multiple Kinects
- Applications where multiple Kinects can be used
- Summary
- 11. Putting Things Together
- Taking Kinect to the Cloud
- Remotely using the Kinect with Windows Phone
- Using Kinect with a Netduino microcontroller
- Augmented reality applications
- Working with face tracking
- Working with XNA and a 3D avatar
- Summary
- Index
The image frames are a type of the ColorImageFrame class. Depending on the type of images you are using, the sensor will return the image frame. To get any color frame from the sensor, you need to either subscribe to the event handler or explicitly send the request to the sensor to send a frame:
There are two ways of capturing the same in your application:
- Event model
- Polling model
Using the event model, the Kinect sensor sends the frame to the application whenever a new frame is captured by the sensor. For that you need to subscribe to the specific event handler using your code, where you need to process the incoming frames. Before subscribing to the event, you must tell the SDK the color type and resolution of image streams you are looking for. Once it's subscribed to, the sensor will send the data continuously unless you disable and unsubscribe the channel, or stop the sensor.
The polling model is an on-demand model where you need to send a request to the sensor whenever there is a necessity to get an image frame. For the polling model, you have to pass the time interval after which the sensor will return the image frame.
Note
If there is any open color channel and there is no subscriber event on that channel, the SDK will automatically pick up the existing channel to get an image frame. If there is no channel open, the SDK will create a new channel to get the frame from the sensor. You can't send a request for images to a channel that is already subscribed by an event handler.
-
No Comment