opengl - 将相机空间中的体素投影到相机视图中，以在 OpenGL 中获得正确的图像坐标

Question

我正在开发一个类似于MobileFusion的应用程序。

在 Sec:6 中，他们通过以下方式将每个体素从相机空间投影到相机视图中：

vec2 = (f_x*(q_x/q_z)+c_x, f_y*(q_y/q/z)+c_y)

I am using a kinect v2 for the images and I found for these parameters:

    float cx = 261.696594;
    float cy = 202.522202;
    float fx = 356.096588;
    float fy = 368.096588;

I want to realize this fusion step with OpenGL. Now my problem is that I have value between [-1,1] and using these parameters for getting the correct image coordinates would yield to way much higher values.

Is there something I do wrong here? Or do I have to normalize these parameters ?