0

I am working on a project which requires me to extract rgb image and corresponding depth map. Since RGB image and depth map are obtained from different sensors they are not synced.

Currently I am trying to modify the example given in pylibfreenect2 official documentation.

    while True:
    frames = listener.waitForNewFrame()

    if enable_rgb:
        color = frames["color"]
        if i%skip_frame ==0 :
            np.save('/home/ashu/DDP/Kinect/RGB_images/rgb_{}'.format(j),color.asarray())
    if enable_depth:
        ir = frames["ir"]
        depth = frames["depth"]

    if enable_rgb and enable_depth:
        registration.apply(color, depth, undistorted, registered)
    elif enable_depth:
        registration.undistortDepth(depth, undistorted)

    if enable_depth:
        cv2.imshow("ir", ir.asarray() / 65535.)
        cv2.imshow("depth", depth.asarray() / 4500.)
        cv2.imshow("undistorted", undistorted.asarray(np.float32) / 4500.)
        if i%skip_frame == 0 :
            np.save('/home/ashu/DDP/Kinect/Depth_images/depth_{}'.format(j),depth.asarray())

    if enable_rgb:
        cv2.imshow("color", cv2.resize(color.asarray(),(int(1920 / 3), int(1080 / 3))))

    if enable_rgb and enable_depth:
        cv2.imshow("registered", registered.asarray(np.uint8))
        xx = registration.getPointXYZ(undistorted, registered , )
        if i%skip_frame == 0 :
            np.save('/home/ashu/DDP/Kinect/registered/reg_{}'.format(j) ,registered.asarray(np.uint8))
            j += 1
    listener.release(frames)
    i+=1
    key = cv2.waitKey(delay=1)
    if key == ord('q'):
        break

device.stop()
device.close()

sys.exit(0)

As it can be seen in the code that I am trying to save rgb image and depth map as .npy file after some fixed number of frames but since both are of different dimensions, they are of no use. Is short my objective is to extract rgb and depth image of same dimensions (424,512)

Ashutosh Singh
  • 1
  • 2
  • 1
    Please take the time to read this: stackoverflow.com/help/how-to-ask . It is easier for use to help when we are able to understand the problem. – Christian Will Nov 15 '17 at 03:13
  • You will not get a solution here if you don't show any own efforts. so saying it would be great to get it in pylibfreenect2 is pointless. shed some light on terms like "registered image". Well you could easily know that after a 5 second web research. enter registered image kinect into google and you will find that it is an image where rgb and depth values are aligned. – Piglet Nov 15 '17 at 07:13
  • Thank you @ChristianWill I have now edited my question suitably. – Ashutosh Singh Nov 16 '17 at 16:55

0 Answers0