ColorFrame to ndarray (Question) about pylibfreenect2 HOT 3 CLOSED

I suspect you are getting dimension mismatch? I'm getting color frame as 4-channel data (whose shape is (1080, 1920, 4)) and can subtract mean pixel value:

>>> im = color.asarray()
>>> im.__class__
<class 'numpy.ndarray'>
>>> im.shape
(1080, 1920, 4)
>>> PIXEL_MEANS = np.mean(im, (0,1))
>>> PIXEL_MEANS.shape
(4,)
>>> PIXEL_MEANS
array([ 132.62429784,  137.5282745 ,  135.42699171,  255.        ])
>>> im - PIXEL_MEANS
array([[[  49.37570216,   44.4717255 ,   45.57300829,    0.        ],
        [  49.37570216,   44.4717255 ,   46.57300829,    0.        ],
        [  48.37570216,   44.4717255 ,   45.57300829,    0.        ],
...

subtracting alpha-channel is not a good idea though (just a demo).

from pylibfreenect2.

@r9y9 Thank you for your response. You made me realize the difference in number of channels, hence I was able to accomplish what I wanted using the OpenCV method cvtColor as in:
newImage = cv2.cvtColor(im, cv2.COLOR_RGBA2RGB)
Have a good one.

from pylibfreenect2.

@r9y9 Hi, Just in case, what do you recommend to do if there is a need to have just 3 channels (RGB)? Is it the only one solution to subtract the alpha-channel? If yes, how to do this in package?

from pylibfreenect2.