Kinect Fusion和关节 [英] Kinect Fusion and joints

问题描述

嗨！

有谁知道如何正确地将关节位置投射到kinect融合体？

Does anyone know how to correctly project joint positions to the kinect fusion volume?

在我的代码中，我为每个关节创建网格，并根据它们的坐标将它们添加到视口中，但它们与融合重建不一致。

In my code, I create meshes for each joint and add them to the viewport based on their coordinates, but they don't align well with the fusion reconstruction.

当使用BodyFrame和Joint位置时，它们在使用坐标映射器时似乎非常准确地对齐。

When using the BodyFrame and Joint positions, they seem to align very accurately when using the coordinatemapper.

我猜有一些方法可以将关节位置转换为"融合位置" ; （类似于坐标映射器），但是如何？

I am guessing there is some way to convert a joint position to a "fusion position" (similar to the coordinate mapper), but how?

搜索此主题，我找不到任何示例或解释（我能理解）。

Searching for this topic, I could not find any examples or explanations (that I could understand).

我们非常感谢任何帮助。

Any help would be greatly appreciated.

推荐答案

Fusion卷不与深度空间坐标系共享原点。如果您查看fusion explorer示例并勾选"显示音量"并取消勾选"Kinect视图"按钮，您可以看到两个坐标系以及它们如何通过音量移动Kinect
。幸运的是，Fusion为您提供了两者之间的转换作为cameraToWorld Matrix4。因此，您可以将您的骨点放在深度空间中，乘以camerToWorld Matix4，以获得世界空间中的点。我认为。

The Fusion volume does not share an origin with the depth space coordinate system. If you look at the fusion explorer sample and tick the 'show volume' and untick the 'Kinect View' button you can see two coordinate systems and how they change as you move the Kinect through the volume. Fortunately, Fusion give you the transformation between the two as the cameraToWorld Matrix4. So you can take your bone point in depth space, multiplied by the camerToWorld Matix4, to get the point in world space. I think.

如果您仍然在后台运行Fusion算法，那么将通过让人在音量中移动来更改cameraToWorld Matrix4。你需要在静态音量中将其冻结，然后开始进入场景中的
的非刚性动作。

If you are still running the Fusion algorithm in the background then the cameraToWorld Matrix4 will be changed by having the human moving around in the volume. You will want to freeze this in a static volume before you start with non-rigid motions coming into the scene.

这篇关于Kinect Fusion和关节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！