我想做一些类似于Johhny Lee在他的Wii头部追踪中所做的事情 http://www.youtube.com/watch?v=Jd3-eiid-Uw&feature=player_embedded
但我想使用Kinect。由于微软的sdk暴露了骨骼关节,我曾希望我能够用它来获得头部位置。问题是我想用我的台式电脑及其显示器来做这件事。如果我将Kinect传感器放在我的显示器旁边并坐在桌子旁边。几乎只有我的头部和颈部对于传感器是可见的,因此骨骼跟踪不会在我的头部位置上拾取。
是否有人熟悉使用Kinect的头部跟踪项目?最好是在C#
答案 0 :(得分:2)
我认为对于此应用程序,您无法使用任何框架(如Microsoft的SDK或OpenNI)提供的骨架跟踪。
我建议通过对原始深度数据应用深度阈值来对用户的头部进行分段。这应该导致背景减少。我认为已有的方法可以做到这一点。
作为第二步,您希望在分段用户中拥有类似轴的东西。最简单的方法是使用opencv fitEllipse。返回椭圆的主轴与深度信息结合使用,可以得到此轴。
此方法仅在大多数分段点属于用户头时才有效。如果你离得更远,你必须考虑一种只分割头部的方法。椭圆拟合应始终有效。
答案 1 :(得分:1)
Windows SDK的官方Kinect存在限制,它与XBox和XDK提供的指导一致,因为您需要距传感器1.2米到3.5米才能使用Kinect传感器。实际上,在OpenNI / NITE库等备用SDK中可以减少此限制,从而可以检测更接近传感器的骨架/对象。
你在skeltal输入时也会遇到的问题是它只会检测头部与骨骼成比例的位置,但如果你将头部左右旋转则不会。为了达到这个目的,你不会使用原始深度流和一些关于物体识别的智能,这有点复杂。
在过去,我使用过这个商业.NET API,它使用网络摄像头来跟踪头部动作,并实现您的目标:http://luxand.com/facesdk/index2.php
答案 2 :(得分:1)
您不需要kinect来跟踪您的头部位置。通过使用面部跟踪,您可以使用普通相机和openCV执行相同的操作。
这里显示的简单示例:http://vimeo.com/19464641
在视频中,我使用openCV来跟踪我的脸(你几乎看不到角落,但红点表示我的脸部位置)。
答案 3 :(得分:1)
查看关于此类主题的第9频道tutorials。你会去参加Skeletal Fundamentals Video。但如果你想节省时间,这里有一些代码 的 XAML 强>
<Window x:Class="SkeletalTracking.MainWindow"
xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
Title="MainWindow" Height="600" Width="800" Loaded="Window_Loaded"
xmlns:my="clr-namespace:Microsoft.Samples.Kinect.WpfViewers;assembly=Microsoft.Samples.Kinect.WpfViewers"
Closing="Window_Closing" WindowState="Maximized">
<Canvas Name="MainCanvas">
<my:KinectColorViewer Canvas.Left="0" Canvas.Top="0" Width="640" Height="480" Name="kinectColorViewer1"
Kinect="{Binding ElementName=kinectSensorChooser1, Path=Kinect}" />
<my:KinectSensorChooser Canvas.Left="250" Canvas.Top="380" Name="kinectSensorChooser1" Width="328" />
<Image Canvas.Left="66" Canvas.Top="90" Height="87" Name="headImage" Stretch="Fill" Width="84" Source="/SkeletalTracking;component/c4f-color.png" />
</Canvas>
内部代码
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Windows;
using System.Windows.Controls;
using System.Windows.Data;
using System.Windows.Documents;
using System.Windows.Input;
using System.Windows.Media;
using System.Windows.Media.Imaging;
using System.Windows.Navigation;
using System.Windows.Shapes;
using Microsoft.Kinect;
using Coding4Fun.Kinect.Wpf;
namespace SkeletalTracking
{
/// <summary>
/// Interaction logic for MainWindow.xaml
/// </summary>
public partial class MainWindow : Window
{
public MainWindow()
{
InitializeComponent();
}
bool closing = false;
const int skeletonCount = 6;
Skeleton[] allSkeletons = new Skeleton[skeletonCount];
private void Window_Loaded(object sender, RoutedEventArgs e)
{
kinectSensorChooser1.KinectSensorChanged += new DependencyPropertyChangedEventHandler(kinectSensorChooser1_KinectSensorChanged);
}
void kinectSensorChooser1_KinectSensorChanged(object sender, DependencyPropertyChangedEventArgs e)
{
KinectSensor old = (KinectSensor)e.OldValue;
StopKinect(old);
KinectSensor sensor = (KinectSensor)e.NewValue;
if (sensor == null)
{
return;
}
var parameters = new TransformSmoothParameters
{
Smoothing = 0.3f,
Correction = 0.0f,
Prediction = 0.0f,
JitterRadius = 1.0f,
MaxDeviationRadius = 0.5f
};
//sensor.SkeletonStream.Enable(parameters);
sensor.SkeletonStream.Enable();
sensor.AllFramesReady += new EventHandler<AllFramesReadyEventArgs>(sensor_AllFramesReady);
sensor.DepthStream.Enable(DepthImageFormat.Resolution640x480Fps30);
sensor.ColorStream.Enable(ColorImageFormat.RgbResolution640x480Fps30);
try
{
sensor.Start();
}
catch (System.IO.IOException)
{
kinectSensorChooser1.AppConflictOccurred();
}
}
void sensor_AllFramesReady(object sender, AllFramesReadyEventArgs e)
{
if (closing)
{
return;
}
//Get a skeleton
Skeleton first = GetFirstSkeleton(e);
if (first == null)
{
return;
}
//set scaled position
ScalePosition(headImage, first.Joints[JointType.Head]);
//ScalePosition(leftEllipse, first.Joints[JointType.HandLeft]);
//ScalePosition(rightEllipse, first.Joints[JointType.HandRight]);
GetCameraPoint(first, e);
}
void GetCameraPoint(Skeleton first, AllFramesReadyEventArgs e)
{
using (DepthImageFrame depth = e.OpenDepthImageFrame())
{
if (depth == null ||
kinectSensorChooser1.Kinect == null)
{
return;
}
//Map a joint location to a point on the depth map
//head
DepthImagePoint headDepthPoint =
depth.MapFromSkeletonPoint(first.Joints[JointType.Head].Position);
//left hand
DepthImagePoint leftDepthPoint =
depth.MapFromSkeletonPoint(first.Joints[JointType.HandLeft].Position);
//right hand
DepthImagePoint rightDepthPoint =
depth.MapFromSkeletonPoint(first.Joints[JointType.HandRight].Position);
//Map a depth point to a point on the color image
//head
ColorImagePoint headColorPoint =
depth.MapToColorImagePoint(headDepthPoint.X, headDepthPoint.Y,
ColorImageFormat.RgbResolution640x480Fps30);
//left hand
ColorImagePoint leftColorPoint =
depth.MapToColorImagePoint(leftDepthPoint.X, leftDepthPoint.Y,
ColorImageFormat.RgbResolution640x480Fps30);
//right hand
ColorImagePoint rightColorPoint =
depth.MapToColorImagePoint(rightDepthPoint.X, rightDepthPoint.Y,
ColorImageFormat.RgbResolution640x480Fps30);
//Set location
CameraPosition(headImage, headColorPoint);
//CameraPosition(leftEllipse, leftColorPoint);
//CameraPosition(rightEllipse, rightColorPoint);
}
}
Skeleton GetFirstSkeleton(AllFramesReadyEventArgs e)
{
using (SkeletonFrame skeletonFrameData = e.OpenSkeletonFrame())
{
if (skeletonFrameData == null)
{
return null;
}
skeletonFrameData.CopySkeletonDataTo(allSkeletons);
//get the first tracked skeleton
Skeleton first = (from s in allSkeletons
where s.TrackingState == SkeletonTrackingState.Tracked
select s).FirstOrDefault();
return first;
}
}
private void StopKinect(KinectSensor sensor)
{
if (sensor != null)
{
if (sensor.IsRunning)
{
//stop sensor
sensor.Stop();
//stop audio if not null
if (sensor.AudioSource != null)
{
sensor.AudioSource.Stop();
}
}
}
}
private void CameraPosition(FrameworkElement element, ColorImagePoint point)
{
//Divide by 2 for width and height so point is right in the middle
// instead of in top/left corner
Canvas.SetLeft(element, point.X - element.Width / 2);
Canvas.SetTop(element, point.Y - element.Height / 2);
}
private void ScalePosition(FrameworkElement element, Joint joint)
{
//convert the value to X/Y
//Joint scaledJoint = joint.ScaleTo(1280, 720);
//convert & scale (.3 = means 1/3 of joint distance)
//Joint scaledJoint = joint.ScaleTo(1280, 720, .3f, .3f);
Canvas.SetLeft(element, scaledJoint.Position.X);
Canvas.SetTop(element, scaledJoint.Position.Y);
}
private void Window_Closing(object sender, System.ComponentModel.CancelEventArgs e)
{
closing = true;
StopKinect(kinectSensorChooser1.Kinect);
}
}
}
我个人会建议观看视频,因为他们会解释所有内容。祝你的项目好运!
答案 4 :(得分:-1)
我建议将Aforge.net与Microsoft XNA Framework一起使用,或仅使用Aforge.net。你需要自己做一些开发。我也在使用C#进行类似的事情。我想你将无法找到一个完整的开箱即用的例子。还没有人做过这件事。 (如果我错了,请纠正我。)