Ask Your Question

Getting clear kinect depth data

asked 2014-11-23 07:36:40 -0500

Aylin COŞKUN gravatar image

updated 2014-11-23 07:37:08 -0500

I am using ros node and getting topic with

sub = nh.subscribe("/camera/depth/image", 1, &Subscribe_Depth::callback, this);

and convert to opencv

cv_ptr = cv_bridge::toCvCopy(im, sensor_msgs::image_encodings::TYPE_32FC1);

I am getting kinect depth data but it is too noisy and shaded image. How do ı clean them from image.Is opencv is good way? How else can I do? Any link or code? (I study on ubuntu and use c++ language )


edit retag flag offensive close merge delete

1 Answer

Sort by » oldest newest most voted

answered 2014-11-24 03:03:16 -0500

Wolf gravatar image

Well, there is no magic "get all noise and shades away" method for kinect data. Both are basically caused by the sensing principle, so in some way somebody has to cope with them. For reducing shades maybe inpainting may help ; for reducing noise maybe gaussian blur may help, ; ; However, in both cases you have to be very careful that you to not "blur away" real features in the image so that it may "look nicer" but you really have lost information. If you want to do some processing on the image, maybe you should try to accept the image has contains some noise and try to use the shades as features....

edit flag offensive delete link more


I will look at them. when ı use opencv the range of kinect is 5 m but my view is not detected after about 1.5 m. I have searched ros with kinect, ı found rosrun image_view image_view image:=/camera/depth/image . This line gives good result. Can I get results like this?

Aylin COŞKUN gravatar image Aylin COŞKUN  ( 2014-11-24 10:58:59 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower


Asked: 2014-11-23 07:36:40 -0500

Seen: 1,066 times

Last updated: Nov 24 '14