You can use simple computer vision algorithms to find circles and/or colours in the RGB image. OpenCV has a lot of examples to do that. Then, after knowing the 2D position of the ball in the RGB image you can look up the pixels in the depth image you you will get the position of the ball in space. Depending on how precise you need it the two images needs to be calibrated.

You can try PCL (point cloud library) for that. If you have 3D point cloud of the scene you can use model segmentation available in PCL for locating the 3D co-ordinates of the ball.

Danish Khan good idea! But does that work in real time?

Dmitry Alexandrovsky PCL have grabbers for both Kinect V1 and V2, Point clouds can be acquired in real time from Kinect using PCL. Considering most PCL algorithms are computationally efficient, i think good fps rate can be achieved.

Can anyone guide me how i to take multiple images from single kinect v1 and how to use them in PCL. Danish Khan @

Maria Komal what do you mean by images, depth/3D point cloud or RGB image. Please be more specific.

BTW, below link is the Kinect V1 grabber for PCL.

https://gist.github.com/UnaNancyOwen/f7cae236b16ac308155b

Thanks for the recommendation, as I am new with kinect v1 so I 'm little bit confuse.

Actually I want to collect dataset of any object from kinect v1. By using point cloud.

I have also download pcl library but how to use it I don't know.

If you just want to use pcl for collecting point cloud data from Kinect, use KScan3D instead it has a simple GUI and you can easily get the 3D point cloud from Kinect.

https://kscan3d.software.informer.com/

Okay. I have one more question when I 'm taking images from kinect. The format of that images or video is .XED now I don't know how to use that image for point cloud.