16UC1 Format Support?

[5730289021-NN]

Greeting everyone,

When I run Intel Realsense D435i on Ubuntu 18.04, I've got these message when I try to access to depth sensor data aka. /camera/depth/image_rect_raw

[ INFO] [1589436848.180278750]: Handling Request: /stream_viewer?topic=/camera/depth/image_rect_raw
[ INFO] [1589436848.199399190]: Handling Request: /stream?topic=/camera/depth/image_rect_raw
[ERROR] [1589436848.323809148]: cv_bridge exception: [16UC1] is not a color format. but [bgr8] is. The conversion does not make sense

I'm not sure whether the 16UC1 support is implemented already.

Does anyone get this problem like this?
Thanks

[jveitchmichaelis]

Have you tried mono16? That ought to work. See the list of names here:

http://wiki.ros.org/cv_bridge/Tutorials/UsingCvBridgeToConvertBetweenROSImagesAndOpenCVImages#cv_bridge.2FTutorials.2FUsingCvBridgeCppHydro.Converting_ROS_image_messages_to_OpenCV_images

The video server won't do min/max scaling of the intensities in the video though, so you may end up with a very dark stream.

[liamabcxyz]

Have you solved the problem? I met the same problem. @5730289021-NN

[jveitchmichaelis]

@liamabcxyz did you try mono16 as well? ROS can definitely handle 16-bit images. Or you can do your own scaling and then feed the video server an 8-bit image. If you need to actually use the data though, you should listen on the topic.

[liamabcxyz]

Hi, @jveitchmichaelis Thanks for your reply! My goal is to display depth image("/camera/depth/image_rect_raw") on webpage. But I don't know how to change 16uc1 to mono16 directly. Thus, my current plan is to send sensor_msgs/Image.msg to webpage, use javascript to display the image. Thanks! I'm looking forward to your reply!

[jveitchmichaelis]

@liamabcxyz This is set in cv_bridge so all you need to do is pass in your 16 bit image (i.e. a cv::Mat) and declare it as mono16. It depends if you're using Python or C++, but essentially you set the encoding flag when you convert your OpenCV Mat. You need to do this at the level of your camera publisher.

Possibly modify it here:

web_video_server/src/image_streamer.cpp

Line 109 in dfc4bf0

img = cv_bridge::toCvCopy(msg, "bgr8")->image;

e.g. add an additional condition to check for 16 bit images. Or, just rescale your 16-bit depth image to an 8-bit range.

else if(msg->encoding.find("16") != std::string::npos){
      img = cv_bridge::toCvCopy(msg, "mono16")->image;
}else{
      // Convert to OpenCV native BGR color
      img = cv_bridge::toCvCopy(msg, "bgr8")->image;
}

see http://wiki.ros.org/cv_bridge/Tutorials/ConvertingBetweenROSImagesAndOpenCVImagesPython

[liamabcxyz]

Hi, @jveitchmichaelis Thanks again! With your help, I write a node to convert 16uc1 to bgr8 with python. It works!

  import roslib
  roslib.load_manifest('converter_depth_image')
  import sys
  import rospy
  import cv2
  import numpy as np
  from std_msgs.msg import String
  from sensor_msgs.msg import Image
  from cv_bridge import CvBridge, CvBridgeError

  class image_converter:

    def __init__(self):
      self.image_pub = rospy.Publisher("image_topic_3",Image)

      self.bridge = CvBridge()
      self.image_sub = rospy.Subscriber("camera/depth/image_rect_raw",Image,self.callback)

    def callback(self,data):

      try:
        data.encoding='mono16'
        cv_image = self.bridge.imgmsg_to_cv2(data, "mono16")
      except CvBridgeError as e:
        print(e)

      img_n = cv2.normalize(src=cv_image, dst=None, alpha=0, beta=255, norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_8U)
      im_color = cv2.cvtColor(img_n, cv2.COLOR_GRAY2BGR)


      try:
        self.image_pub.publish(self.bridge.cv2_to_imgmsg(im_color, "bgr8"))
      except CvBridgeError as e:
        print(e)

  def main(args):
    rospy.init_node('image_converter', anonymous=True)
    ic = image_converter()

    try:
      rospy.spin()
    except KeyboardInterrupt:
      print("Shutting down")

  if __name__ == '__main__':
      main(sys.argv)

[RalphCodesTheInternet]

How was the latency with this approach?

Have you by any change figured out how to decode 16UC1 messages sent from rosbridge?

[zaksel]

Hi, i also use a Realsense camera with the provided ros wrapper and also stumbled on this error message.

I got the depth stream working by adjusting the image_streamer.cpp like mentioned by @jveitchmichaelis.
On the ros wrapper of the realsense i changed the encoding to mono16 by changing the following lines in base_realsense_node.cpp:
Line 96
_encoding[RS2_STREAM_DEPTH] = sensor_msgs::image_encodings::TYPE_16UC1; // ROS message type
_encoding[RS2_STREAM_DEPTH] = sensor_msgs::image_encodings::MONO16; // ROS message type
and also Line119
_depth_aligned_encoding[RS2_STREAM_COLOR] = sensor_msgs::image_encodings::TYPE_16UC1;
_depth_aligned_encoding[RS2_STREAM_COLOR] = sensor_msgs::image_encodings::MONO16;

Maybe this can be helpful for others :)

[StevenqrYang]

encount error like this when using your decoding function :

[ERROR] [1687136715.285666]: bad callback: <bound method detect.read_depimg of <__main__.detect object at 0x7f0c7142e8>>
Traceback (most recent call last):
  File "/opt/ros/melodic/lib/python2.7/dist-packages/rospy/topics.py", line 750, in _invoke_callback
    cb(msg)
  File "rstest_ros.py", line 281, in read_depimg
    self.depth_image = bridge.imgmsg_to_cv2(data, "mono16")
  File "/home/jetson/catkin_cvbridge_ws/src/vision_opencv/cv_bridge/python/cv_bridge/core.py", line 168, in imgmsg_to_cv2
    dtype=dtype, buffer=img_msg.data)
TypeError: buffer is too small for requested array

As the data of this topic is stored as a list.