2012:Shooter and Collector Camera: Difference between revisions

From 1511Wookiee
Jump to navigationJump to search
No edit summary
 
(23 intermediate revisions by 2 users not shown)
Line 1: Line 1:
=== <span style="font-size: larger;">'''1/15: tested white and red led reflections'''</span> ===
== <span style="font-size: larger">'''Todo'''</span> ==
*run 1/21 images through image library to determine best algorithm to use (hull vs edge)
*use color or grayscale camera image (will depend on algorithm used)
*determine how fast NI image library is (if using edge detect, might want to right that part ourselves)
*test green and white led ring when they come in
*...
 
== <span style="font-size: larger">'''Resources'''</span> ==
*2012 Vision Info: http://firstforge.wpi.edu/sf/docman/do/downloadDocument/projects.wpilib/docman.root/doc1302
*Hull thread: http://www.chiefdelphi.com/forums/showthread.php?threadid=100176
*Camera settings: [http://penfieldrobotics.com/wiki/index.php?title=File:axis_camera_settings1.jpg 1]; [http://penfieldrobotics.com/wiki/index.php?title=File:axis_camera_settings2.jpg 2] NOTE: Automatic exposure should work as well
 
== <span style="font-size: larger">'''Notebook'''</span> ==
==== <span style="font-size: larger">'''1/15: tested white and red led reflections'''</span> ====


*Target: 26 feet from camera; OSB; tape is black duct tape (note: not enough tape to do inside completely)
*Target: 26 feet from camera; OSB; tape is black duct tape (note: not enough tape to do inside completely)
*Light Sources: 1) Round light with several red leds (something from a prior year); 2) Mini-led flashlight (red from toolbox), has 5 wihtie leds
*Light Sources: 1) Round light with several red leds (something from a prior year); 2) Mini-led flashlight (red from toolbox), has 5 wihtie leds
*Results: The red led gave off a little red cast (each time we connected power, a different number of leds lit up). The white leds reflected the best. Concerned the white leds might be too bright for drivers on other alliance
*Results: The red led gave off a little red cast (each time we connected power, a different number of leds lit up). The white leds reflected the best. Concerned the white leds might be too bright for drivers on other alliance
*Images:
*Images: 1) [http://penfieldrobotics.com/wiki/index.php?title=File:Color005.png No light source]; 2) [http://penfieldrobotics.com/wiki/index.php?title=File:Color004.png Red led light]; 3) [http://penfieldrobotics.com/wiki/index.php?title=File:Color003.png Mini flashlight]
 
 


=== <span style="font-size: larger;">'''1/19: tested Infrared lights reflecting off reflective tape for the camera'''</span> ===
==== <span style="font-size: larger">'''1/19: tested Infrared lights reflecting off reflective tape for the camera'''</span> ====


*Target: about 4 feet from camera (we did not measure it); poly; outer tape is black duct tape and inner tape is blue gaffers tape; in the middle of the target, the plastic protection is on the back; area outside the black tape has the plastic protection on both sides
*Target: about 4 feet from camera (we did not measure it); poly; outer tape is black duct tape and inner tape is blue gaffers tape; in the middle of the target, the plastic protection is on the back; area outside the black tape has the plastic protection on both sides
*Light Sources: 1) infared LED Night light, Model # IR045; 2) Mini-led flashlight (red from toolbox), has 5 wihtie leds
*Light Sources: 1) infared LED Night light, Model # IR045; 2) Mini-led flashlight (red from toolbox), has 5 wihtie leds
*Results: Camera can infrared light, but no reflection on tape (you can also see reflect on the plastic).
*Results: Camera can infrared light, but no reflection on tape (you can also see reflect on the plastic).
*Images: 1) ; 2) ; 3)
*Images: 1) [http://penfieldrobotics.com/wiki/index.php?title=File:Image_no_ir.jpg No light source]; 2) [http://penfieldrobotics.com/wiki/index.php?title=File:Image_with_ir.jpg Infared light]; 3) [http://penfieldrobotics.com/wiki/index.php?title=File:Image_with_white_led.jpg Mini flashlight]
 
 
 
==== <span style="font-size: larger">'''1/21: ran edge detection on images from 1/15 and 1/19'''</span> ====
 
*1/15 Images: 1) [http://penfieldrobotics.com/wiki/index.php?title=File:Color0005.bmp.png No light source]; 2) [http://penfieldrobotics.com/wiki/index.php?title=File:Color004.bmp.png Red led light]; 3) [http://penfieldrobotics.com/wiki/index.php?title=File:Color003.bmp.png Mini flashlight]
*1/21 Images: 1) [http://penfieldrobotics.com/wiki/index.php?title=File:Image_no_ir.jpg.png No light source]; 2) [http://penfieldrobotics.com/wiki/index.php?title=File:Image_with_ir.jpg.png Infared light.png]; 3) [http://penfieldrobotics.com/wiki/index.php?title=File:Image_with_white_led.jpg.png Mini flashlight]
 
==== <span style="font-size: larger">'''1/21: Took lots of pictures of the target:'''</span> ====
 
*<font size="2">From&nbsp;various distances/locations:</font>
**up against Fender<font size="2"></font>
**sideways at Fender
**<font size="2">top of&nbsp;Key</font>
**<font size="2">corner of Key'''&nbsp;'''</font>
**<font size="2">at center Bridge</font>
**<font size="2">at alliance Bridge</font>
*At all three hoop heights
**Taking into account sideways displacement of middle hoops
*Some with light shining on the retro-reflective tape, some without
*Some with mock "hoop", some without
*Some with balls blocking parts of target, some without
*Target always approximately in the center of the picture
**This assumes variable-angle camera
 
==== <span style="font-size: larger">'''1/22: Image processing'''</span> ====
*Converted 1/21 images. The lit reflective tape does improve the edge detection. The balls and hoop do block the target, so this will need to be taken into account.
*2012 Vision Info: http://firstforge.wpi.edu/sf/docman/do/downloadDocument/projects.wpilib/docman.root/doc1302
*LED Samples (33mm inner dia): http://www.superbrightleds.com/cgi-bin/store/index.cgi?action=DispPage&category=ACCENTS&Page2Disp=%2Fspecs%2FAE.htm

Latest revision as of 13:09, 3 March 2012

Todo

  • run 1/21 images through image library to determine best algorithm to use (hull vs edge)
  • use color or grayscale camera image (will depend on algorithm used)
  • determine how fast NI image library is (if using edge detect, might want to right that part ourselves)
  • test green and white led ring when they come in
  • ...

Resources

Notebook

1/15: tested white and red led reflections

  • Target: 26 feet from camera; OSB; tape is black duct tape (note: not enough tape to do inside completely)
  • Light Sources: 1) Round light with several red leds (something from a prior year); 2) Mini-led flashlight (red from toolbox), has 5 wihtie leds
  • Results: The red led gave off a little red cast (each time we connected power, a different number of leds lit up). The white leds reflected the best. Concerned the white leds might be too bright for drivers on other alliance
  • Images: 1) No light source; 2) Red led light; 3) Mini flashlight


1/19: tested Infrared lights reflecting off reflective tape for the camera

  • Target: about 4 feet from camera (we did not measure it); poly; outer tape is black duct tape and inner tape is blue gaffers tape; in the middle of the target, the plastic protection is on the back; area outside the black tape has the plastic protection on both sides
  • Light Sources: 1) infared LED Night light, Model # IR045; 2) Mini-led flashlight (red from toolbox), has 5 wihtie leds
  • Results: Camera can infrared light, but no reflection on tape (you can also see reflect on the plastic).
  • Images: 1) No light source; 2) Infared light; 3) Mini flashlight


1/21: ran edge detection on images from 1/15 and 1/19

1/21: Took lots of pictures of the target:

  • From various distances/locations:
    • up against Fender
    • sideways at Fender
    • top of Key
    • corner of Key 
    • at center Bridge
    • at alliance Bridge
  • At all three hoop heights
    • Taking into account sideways displacement of middle hoops
  • Some with light shining on the retro-reflective tape, some without
  • Some with mock "hoop", some without
  • Some with balls blocking parts of target, some without
  • Target always approximately in the center of the picture
    • This assumes variable-angle camera

1/22: Image processing