Jordan.Brenner's profile - activity

From here you can write the video to a file and then upload it.

The function Undistort can be used to display the rectification. Try something like this:

# Left camera
dst1=cv2.undistort(vidL,cameraMatrixLeft,distCoeffsLeft) # apply calibration to left camera
cv2.namedWindow('Left Calibrated',cv2.WINDOW_NORMAL) # resizeable window
cv2.imshow('Left Calibrated',dst1) # display left image
# Right camera
dst2=cv2.undistort(vidR,cameraMatrixRight,distCoeffsRight) # apply calibration to right camera
cv2.namedWindow('Right Calibrated',cv2.WINDOW_NORMAL) # resizeable window
cv2.imshow('Right Calibrated',dst2) # display right image

Cheers, Jordan

What is the cv2.stereoCalibrate: "imageSize" variable format?

I've tried numpy arrays, defining the variable plainly, pixel count (720 x 1280 x 3), etc. but I'm still getting the below error. I can't seem to find any documentation on it. Please help!

Snippet of code:

# Stereo calibration
imageSize = (720,1280)
retval_stereo,cameraMatrixLeft,distCoeffsLeft,cameraMatrixRight,distCoeffsRight,R,T,E,F=cv2.stereoCalibrate(
    objPoints,
    imgPoints1,
    imgPoints2,
    cameraMatrix1,
    distCoeffs1,
    cameraMatrix2,
    distCoeffs2,
    imageSize
    )

Error message:

Traceback (most recent call last):
  File "CalibrationOpenCV_TEST.py", line 78, in <module>
    imageSize
SystemError: new style getargs format but argument is not a tuple

Comment (2/2) It opened up a new window labeled Left with nothing shown (white background). The window doesn't respond. After a few seconds the OS takes over requiring the window to be forced to quit.

When the print statement is un-commented the ret remains true until the script is killed.

Comment (1/2) Is there any reason this shouldn't work?

#!usr/bin/python
import cv2
import numpy as np

# script imports Zed camera feed and splits the two feeds into left and right frames

# Capture video on camera device 1
vid = cv2.VideoCapture(1)

while(True):
# Capture frame-by-frame
    ret, img =  vid.read() # please check if ret==True. only fools don't.
    #print "Check if true: %d" % ret
    imgLeft  = img[1:1242, 1:2208]
    imgRight = img[1:1242, 2208:4416]
# Display the resulting frame
    # left
    #cv2.namedWindow('Left', cv2.WINDOW_NORMAL)
    #cv2.resizeWindow('Left', 1920, 1080)
    cv2.imshow('Left', imgLeft)
    # right
    #cv2.namedWindow('Right', cv2.WINDOW_NORMAL)
    #cv2.resizeWindow('Right', 1920, 1080)
    cv2.imshow('Right', imgRight)

# End Process
cv2.waitKey(0)
cv2.destroyAllWindows()

Thanks for your help.

I included the code you provided as follows: #!usr/bin/python import cv2 import numpy as np

# Capture video on camera device 1
vid = cv2.VideoCapture(1)

while(True):
# Capture frame-by-frame
    img, ret =  vid.read() # please check if ret==True. only fools don't.
    imgLeft  = img[1:1080, 1:1920]
    imgRight = img[1:1080, 1920:3840]
# Display the resulting frame
    # left
    cv2.namedWindow('Left', cv2.WINDOW_NORMAL)
    cv2.resizeWindow('Left', 1920, 1080)
    cv2.imshow('Left', imgLeft)
    # right
    cv2.namedWindow('Right', cv2.WINDOW_NORMAL)
    cv2.resizeWindow('Right', 1920, 1080)
    cv2.imshow('Right', imgRight)

Error Message: imgLeft = img[1:1080, 1:1920] TypeError: 'bool' object has no attribute '__getitem__'

I will when I get the camera running. That was just an attempted quick fix.

Background: I was given a Zed Camera with the task of using it as a stereo camera. StereoLabs provided a proprietary software which requires CUDA (NVIDIA hardware-dependent software package) to run. I found a few tutorials for making and calibrating a stereo camera (example). I need to reformat the Zed camera's video feed in order to use this method.

Goal: I'm trying to set up my zed camera as a stereo camera. The camera takes images which look like this: (see Fig. 1) and using the below script I separated the static image into two halves: (see Figs 2,3)

Script:

#!usr/bin/python
import cv2
import numpy as np

# Import image
img = cv2.imread('TestPicture3.jpg')

# split image (indexing)
image_left = img[1:1080, 1:1920]
image_right = img[1:1080, 1920:3840]

# Display left and right images
cv2.namedWindow('Left', cv2.WINDOW_NORMAL)
cv2.resizeWindow('Left', 1920, 1080)
cv2.imshow('Left', image_left)

cv2.namedWindow('Right', cv2.WINDOW_NORMAL)
cv2.resizeWindow('Right', 1920, 1080)
cv2.imshow('Right', image_right)

cv2.waitKey(0)

# export images
cv2.imwrite('TestPic_left.jpg', image_left)
cv2.imwrite('TestPic_right.jpg', image_right)

cv2.destroyAllWindows()

I want to bisect the video feed in order to capture left and right video. Below is the script I've written to attempt the bifurcation, but it hasn't worked. What should I do?

#!usr/bin/python
import cv2
import numpy as np

# script imports Zed camera feed and splits the two feeds into left and right frames

# Capture video on camera device 1
vid = cv2.VideoCapture(1)

# Split video feed on left side
imgLeft = vid.read[1:1080, 1:1920]

# Split video feed on right side
imgRight = vid.read[1:1080, 1920:3840]

while(True):
# Capture frame-by-frame
    ret, frame = vid.read()
# Display the resulting frame
    # left
    cv2.namedWindow('Left', cv2.WINDOW_NORMAL)
    cv2.resizeWindow('Left', 2208, 1242)
    cv2.imshow('Left', imgLeft)
    # right
    cv2.namedWindow('Right', cv2.WINDOW_NORMAL)
    cv2.resizeWindow('Right', 2208, 1242)
    cv2.imshow('Right', imgRight)

# End Process
cv2.waitKey(0)
cv2.destroyAllWindows()

The resulting error is:

Traceback (most recent call last):
  File "Split_VideoCapture_Test.py", line 11, in <module>
    imgLeft = vid.read[1:1080, 1:1920]
TypeError: 'builtin_function_or_method' object has no attribute '__getitem__'

Thank in advance for the assistance.

Cheers, Jordan