Early development experience of Kinect for Windows SDK (3) skeleton tracking

ArticleDirectory

• Initialization code
• Add code
• Written to the end

Author: Ma Ning

The development of our Kinect SDK is getting better and better. Skeleton tracking is the core technology of Kinect. With this technology, many interesting functions can be implemented.

First, let's take a look at the implementation of bone tracing. Kinect can track up to 20 bone points, and currently only the human body can be tracked, other objects or animals can't do anything. This section describes the distribution of the skeleton points of Kinect:

Initialization Code

Next, let's take a look at how the code for bone tracing is compiled. First, we need to create a new Visual C # project called "skeletontracking" and add the KinectProgramFor the work of the coding4fun assembly and coding4fun assembly, refer to the content in the previous article "experience the development of the Kinect for Windows SDK (ii) operating on camera", which will not be repeated here.

First, create a runtime object. during initialization, specify the runtimeoptions of useskeletaltracking, and then add the handler in the skeletonframeready event.

 runtime Nui;  private   void  window_loaded ( Object  sender, routedeventargs e) {Nui =  New  Runtime (); Nui. initialize (runtimeoptions. useskeletaltracking); Nui. skeletonframeready + =  New  eventhandler 
  
    (nui_skeletonframeready);} 
  

Next, the event processing function when the form is closed:

 
Private VoidWindow_closed (ObjectSender, eventargs e) {Nui. uninitialize ();}
 

At this time, add a breakpoint in the null nui_skeletonframeready event handler:

 
VoidNui_skeletonframeready (ObjectSender, skeletonframereadyeventargs e ){}
 

The skeletonframeready event is triggered when you connect to the Kinect device correctly and stand in front of it so that the player can properly recognize the human body.

We can see the returned event processing parameters, of which skeletonframe and skeletons are important.

Add code

Next, we will prepare the WPF interface, and add five balls on the interface to track the positions of the head, hands, and knees. In mainpage. XAML, add the following code:

 < Canvas   Name  = "Maincanvas"  >  <  Ellipse   Canvas . Left  = "0"   Canvas . Top  = "0"   Height  = "50"   Name  = "Headellipse"   Stroke  = "Black"   Width  = "50"   Fill = "Orange"   />  <  Ellipse   Canvas . Left  = "50"   Canvas . Top  = "0"   Height  = "50"   Name  = "Rightellipse"   Stroke  = "Black"   Width  = "50"   Fill  = "Slategray"  />  <  Ellipse   Canvas . Left  = "100"   Canvas . Top  = "0"   Fill  = "Springgreen"   Height  = "50"   Name  = "Leftellipse"   Stroke  = "Black"   Width  = "50"   /> <  Ellipse   Canvas . Left  = "150"   Canvas . Top  = "0"   Height  = "50"   Name  = "Kneerightellipse"   Stroke  = "Black"   Width  = "50"   Fill  = "Salmon"   />  < Ellipse   Canvas . Left  = "200"   Canvas . Top  = "0"   Fill  = "Navy"   Height  = "50"   Name  = "Kneeleftellipse"   Stroke  = "Black"   Width  = "50"   />  </  Canvas > 

Then, we add the method to capture skeletondata in the skeletonframeready event handler:

  void  nui_skeletonframeready ( Object  sender, skeletonframereadyeventargs e) {skeletonframe allskeletons = E. skeletonframe;  // get the first tracked skeleton  skeletondata skeleton = (from S  in  allskeletons. skeletons  where  S. trackingstate = skeletontrackingstate. trackedselect S ). firstordefault () ;}

We use LINQ to obtain the skeletondata whose trackingstate is equal to tracked. The joints attribute set of the skeletondata object stores information about all the skeleton points. The information of each bone point is a joint object, where position X, Y, and Z represent three-dimensional positions. The range of X and Y is-1 to 1, while Z is the distance from Kinect to the recognition object.

We can use the following code to scale the joint to a proper ratio:

Joint J = skeleton. joints [jointid. handright]. scaleto (640,480,. 5f,. 5f );
 

The last two parameters are the maximum and minimum values of the original size. The preceding statement is equivalent to extending the range from-0.5 to 0.5 to 0 to 640.

We encapsulate a function to convert the obtained skeletondata to a circle on the screen:

Private VoidSetellipseposition (frameworkelement ellipse, joint) {var scaledjoint = joint. scaleto (640,480 ,. 5f ,. 5f); canvas. setleft (ellipse, scaledjoint. position. x); canvas. settop (ellipse, scaledjoint. position. Y );}
 

Finally, we can handle the skeletonframeready event as follows:

 
VoidNui_skeletonframeready (ObjectSender, skeletonframereadyeventargs e) {skeletonframe allskeletons = E. skeletonframe;// Get the first tracked skeletonSkeletondata skeleton = (from SInAllskeletons. SkeletonsWhereS. trackingstate = skeletontrackingstate. trackedselect S ). firstordefault (); setellipseposition (headellipse, skeleton. joints [jointid. head]); setellipseposition (leftellipse, skeleton. joints [jointid. handleft]); setellipseposition (rightellipse, skeleton. joints [jointid. handright]); setellipseposition (kneeleftellipse, skeleton. joints [jointid. kneeleft]); setellipseposition (kneerightellipse, skeleton. joints [jointid. kneeright]);}
 

Finally, the program runs as follows. It seems that knee recognition still has some problems:

When the program is running, we will find that the ball is beating. to reduce this problem, we need to set the transformsmooth attribute of the skeletonengine to true and specify the transformsmoothparameters parameter, this parameter should also be fine-tuned based on the application's actual situation.

The load function after the code is added. The Code is as follows:

  private   void  window_loaded ( Object  sender, routedeventargs e) {Nui =  New  Runtime (); Nui. initialize (runtimeoptions. useskeletaltracking);  // must set to true and set after call to initialize  Nui. skeletonengine. transformsmooth =  true ;  // use to transform and reduce jitter  var parameters =  New  transformsmoothparameters {smoothing = 0.75f, correction = 0.0f, prediction = 0.0f, jitterradius = 0.05f, maxdeviationradius = 0.04f}; Nui. skeletonengine. smoothparameters = parameters; Nui. skeletonframeready + =  New  eventhandler 
  
    (nui_skeletonframeready);} 
  

Written to the end

Here, we have introduced the best part of "skeleton tracking" in Kinect. You can write some interesting applications. Next, we will introduce the depth data, another core feature of Kinect.

 

Openxlive cup Windows Phone game development competition

Openxlive cup Windows Phone game development competition is jointly developed by openxlive and well-known developers in ChinaCommunity: Devdiv, smart machine network, wpmind, Silverlight silver light China and xNa game world, a Windows Phone game development competition.

Http://www.openxlive.net/posts/news/40