Alignment of data

From zed mini camera,we can get RGB(left image) data, depth data and ptcloud data.

First of all, what kind of numpy matrix do you convert these data into?
In zed-tensorflow, we use the following code transform the left(RGB image) into a unit8 numpy matrix:
def load_image_into_numpy_array(image):
ar = image.get_data()
ar = ar[:, :, 0:3]
(im_height, im_width, channels) = image.get_data().shape
return np.array(ar).reshape((im_height, im_width, 3)).astype(np.uint8)
What kind of numpy matrix can be converted from depth image and point cloud for further processing?

The second question is, can these three kinds of data correspond? I mean, after obtaining the numpy matrix of three kinds of data, how to obtain the depth data corresponding to a pixel in RGB and the three-dimensional position represented by the point cloud?