Stereo Vision and Projected Light

[Klemen]

Hello,

I used 9 pairs of images. Depends on your sensor/lense quality. The images are used (in the first stage) to model the camera centers, focal lenghts, lense distortion. In the next stage the stereo calibration is performed based on the correspondences between both images. Make sure that in the first set of images, the entire grid is visible to both cameras.

Take a look at this link:

http://zone.ni.com/reference/en-XX/help/372916P-01/nivisionconceptsdita/guid-fafab0af-2f6a-4130-ad6c...

Best regards,

K

[Ivan0907]

Hello K,

Are u in Christmas?

Could you provide the calibration grid images to me? I'd like to use your grid image for calibration.

Thank you.

[Ivan0907]

Hello K,

Can I know how you acquire your phone in a setting of 640x480? My camera is "microsoft 1080p lifecam studio", and it always in 1920x1080 when I acquire the 9 pairs of photos, I think it is the main reason I cann't get the calibration data maybe.

B.R

Ivan

[Ivan0907]

Hello K,

Sorry, can I bother you to generate "StereoCalibrationData" by the linked the images from here

==========================================================================

https://drive.google.com/folderview?id=0B8FTOXbKQhb6VWU4SkdDNHc2TkE&usp=sharing

==========================================================================

Thank you for help.

B.R

Ivan Cheng

[Klemen]

Hello,

sorry for the late reply, I have been busy a bit.

I saw your attachment and I have a few major doubts about your calibration:

1. Why are all your images the same? There is no movement of the calibration grid.

2. Your calibration grid is not planar but curved. You should attach your grid to a planar object that you can move within the measuring field of view.

Best regards,

K

[Ivan0907]

Hello K,

Thank you for replying, please see my answer below:

1. Why are all your images the same? There is no movement of the calibration grid.

    Ivan: How you move your calibration grid, in the first capturing the images, I'm not sure how to move the calibration grid.

2. Your calibration grid is not planar but curved. You should attach your grid to a planar object that you can move within the measuring field of view.

    Ivan: Thank you, I will try to set a planar object for capturing image. 

B.R

Ivan

[Klemen]

Hello,

take a look at the NI stereo example:

"Compute Depth Image.vi"

Best regards,

K

[Ivan0907]

Hello K,

Thank you for the information.

I think you were meaning that "Stereo Vision Example"

Basic now I captured the "StereoCalibrationData" after put that paper of calibration grid in a plane surface, also set in 640x480 and move calibration grid during acquisition.

Now my question is during calibration, I cannot get all red points, whether it is ok or not and how can I get all red points in calibrating.

BTW, did you know what are those data in "StereoCalibrationData"? I have opened by .txt, but I don't understand what is that.

Thank you.

B.R

Ivan Cheng

[Klemen]

Hello,

the example I am reffering to is called "Compute Depth Image.vi" (at least for LV 2013). Start Labview->Help->Find Examples->Search->Compute Depth Image.

Looking at your example image, you need to change the parameters for the dot pattern detection -  i.e. you need to tweak the parameters of the dot detection in the subVI "Extract Grid.vi" (sequence 0 in my Calibration.vi).

The output of the calibration file is encoded, but it consists of extrinsic and intrinsic camera parameters. You can read the file using the "Get Binoclar Stereo Calibration Info".

Best regards,

K

[Ivan0907]

Hello K,

Thank you for response.

Due to my Labview is 2012, so I cann't fine the example you saying in this version, even I copying from 2014, whether it could be used in 2012?

In anohter side, I have acquisition photos for calibration again, and also I got the calibration data, but I don't know why the calibration data seems cannot work fine in measurement, Can you help check what's problem with this, the realtive infomration has uploaded in below link for you referring, thank you.

https://drive.google.com/file/d/0B8FTOXbKQhb6TWRWTzdGSzFnQTA/view?usp=sharing

B.R

Ivan Cheng