ipisoft.com
http://forum.ipisoft.com/

Calibration problem dual kinect
http://forum.ipisoft.com/viewtopic.php?f=2&t=10067
Page 1 of 1

Author:  Wylwers31 [ Sat Dec 29, 2018 7:46 am ]
Post subject:  Calibration problem dual kinect

Hello,

could a person explain to me how to do a calibration with 2 kinect correctly? Need a big board?

Thank you

Author:  vmaslov [ Sun Dec 30, 2018 12:57 pm ]
Post subject:  Re: Calibration problem dual kinect

Hi
Actually we recommend to calibrate using a light marker. The procedure is described in detail in our docs.
http://docs.ipisoft.com/User_Guide_for_Multiple_Depth_Sensors_Configuration#Calibration

However, you still can use the board calibration method for dual Kinect setup if you find it more convenient. Board calibration is described here
http://wiki.ipisoft.com/User_Guide_for_Dual_Depth_Sensor_Configuration#Calibration
The recommended board size is 1 square meter or a bit more.

Author:  Greenlaw [ Wed Jan 02, 2019 12:52 pm ]
Post subject:  Re: Calibration problem dual kinect

I've used both calibration methods for Kinect and Kinect 2 setups. Between the two, light calibration is more accurate but board calibration is easier and quicker to perform and (IMO) usually works well enough.

Here are some tips.

Point Light calibration for depth sensors is the same as for video based cameras. Ideally, you want a small Maglite, the type you can uncap the head from to reveal the led bulb. You can then fit the head on the bottom and use it as a base. This configuration called 'candle mode.' The trick with using a point light for calibration is that you want to avoid occluding the light from all Kinect sensors. This is actually a little easier with Kinect since you're probably recording with fewer devices than with, say, PS3 Eye.

One important consideration: You must to set Recorder to record the additional RGB data because the light is not visible in the depth data.

Alternatively, you can use the glowing ball end of a PS Move controller for light tracking.

Flat Plane calibration can be done with a board measuring at least 1 x 1 meter. I prefer to use a 30 x 40 inch sheet of foam core...it's large enough and rigid enough for good calibration and it's very light weight. You can purchase foam core board at any art, craft, or office supply store.

There are two ways to use the board.

If both sensors are on the same side of the capture space, you stand with the board held out in front of you and move back and forth. Take actual steps toward and away from the sensors because you want to avoid bringing the board close to your body and record bad data.

If the sensors are on opposite sides of the space (facing each other but not directly facing,) then you want to hold the board out to the side and walk back and forth between the sensors. Because the foam core board is light weight, I can usually hold it out with one arm.

It's important to hold the board away from your body. In general, you want to prevent Recorder from merging your body's depth data with the board's.

Note that the board method does not require the RGB data so you can disable it if you wish. Disabling RGB results in smaller capture files that merge more quickly over the network. (Which is why I sometimes still use the board method.) Also, if a computer is not very fast, disabling RGB capture may be necessary. (And, in this situation, light calibration will not be an option.)

Hope this helps.

Author:  Greenlaw [ Wed Jan 02, 2019 1:02 pm ]
Post subject:  Re: Calibration problem dual kinect

Last thoughts:

Board calibration is for two sensors, but you can optionally use light calibration.

If you have three or more sensors, then you MUST use light calibration.

Recording with a single sensor is 'self calibrating' and doesn't need the light or board.

Page 1 of 1 All times are UTC - 8 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group
http://www.phpbb.com/