如何在Kinect for Windows v2中获取手腕的旋转信息？ [英] How can I get the rotation information of my wrists in Kinect for Windows v2 ?

问题描述

所有

假设我们将人的手腕作为三维空间笛卡尔坐标的坐标原点，那么当手腕自由移动时，手腕是什么坐标系中的旋转信息？ kinect V2可以像V2 SDK浏览器中给出的Face Basics演示中的滚动，俯仰和偏航程度一样描述它吗？b


基于Kinect Studio 2.0在3D视图模型上，它实时给出每个关节的矢量箭头，当我移动时，箭头的方向相应地改变。所以我认为这可以帮助我获得我想要的信息。  
$


以下是我的问题：

1。我可以使用Kinect V2 SDK直接检测和输出这些信息吗？如果没有，是否有任何建议使用SDK输出的数据来获取轮换信息？    

2。如果Q1的答案是肯定的，那么传感器关于手腕上的滚动，俯仰和偏航程度的准确度是多少？

3。我已经阅读了关于body类的Kinect V2 SDK文档，JointOrientations属性描述说"body的联合方向"，我想知道的是基于
的联合方向是什么参考框架？ body.JointOrientations的具体值看起来是什么类型的数据？

Assuming we make human wrist as the coordinate origin point of a three-dimensional space Cartesian coordinate, then when moving the wrist freely, what is the wrist rotation information in the coordinate system? Can kinect V2 describe it just like the degree of roll, pitch and yaw in Face Basics demo given in the V2 SDK Browser?

Based on Kinect Studio 2.0 on 3D View model, it gives a vector arrow at each joint in real time, when I move, the direction of arrows changes correspondingly. So I think this may help me to get the information I want.  

Here are my questions:
1. Could I use Kinect V2 SDK to detect and output this information directly? If not, any suggestion to use the data which SDK output to get rotation information?    
2. If the Q1 answer is yes, what is the accuracy of the sensor about the degree of roll, pitch and yaw on wrist?
3. I have read Kinect V2 SDK documentation about the body class, the JointOrientations property description says that "the joint orientations of the body", what I want to know is the joint orientation based on what frame of reference? The specific value of body.JointOrientations looks what kind of data?

任何建议或指导  将大大
赞赏。 

Any advise or guidance would be greatly appreciated. 

推荐答案

是的，您可以通过JointOrienations从核心机构数据中获取联合轮换信息。它作为四元数值提供，其中Y分量旋转围绕来自父母的方向向量; Z是法线，X是双法线。你
可以查找如何在互联网上转换为欧拉角度。准确度取决于传感器的位置以及当手遮住手腕时可能发生的噪音和干扰因素。您需要实施某种类型的过滤，因为已知价值可能是
在某些情况下会旋转180度。

Yes you can get joint rotation information from the core body data through JointOrienations. It is provided as a quaternion value, where the Y component rotation is around the direction vector from the parent; Z is the normal and X is the bi-normal. You can lookup how to convert to Euler angles on the Internet. The level of accuracy is going to depend on sensor placement and factors of noise and interference that may occur when hands occlude the wrist. You will need to implement some type of filtering as it is known that the value could be 180 rotated in some cases.

方向值基于相机坐标系，而不是来自父级的旋转。您可以通过乘以父级的反四元数（共轭）来创建局部旋转。

The orientation value is based on the camera coordinate system, not the rotation from the parent. You can create a local rotation by multiplying by the inverse quaternion(conjugate) of the parent.

这篇关于如何在Kinect for Windows v2中获取手腕的旋转信息？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！