Revision history [back]

When using findCirclesGrid in python it seems that the ordering of the found circles is inverted. My found circles are currently ordered:

• row wise: bottom to top
• column wise: right to left

According to the documentation (at least how I interpret it) I expected it to be the other way around:

• row wise: top to bottom
• column wise: left to right

Several months ago I used this function in c++ and it seemed to behave in the manner as I expected (top to bottom, left to right).

Am I doing something wrong / interpreting the documentation incorrectly or is this a bug that should be reported?

Below a minimal working example which works on the attached bitmap.

edit: image did not get attached. Posted the image here http://imgur.com/a/ROVss

#!/usr/bin/python

import cv2

import numpy as np

img = cv2.imread('circ_min.bmp', 0)

shape = (5, 8)

blobPar = cv2.SimpleBlobDetector_Params()

blobPar.minArea = 1e4

blobPar.maxArea = 1e5

blobPar.minDistBetweenBlobs = 200

blobDet = cv2.SimpleBlobDetector_create(blobPar)

[found, cc] = cv2.findCirclesGrid(img, shape, None, flags=cv2.CALIB_CB_SYMMETRIC_GRID+cv2.CALIB_CB_CLUSTERING, blobDetector=blobDet)

cv2.namedWindow('src', cv2.WINDOW_NORMAL)

img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)

if found:

print(cc)

for i in cc:

    cv2.circle(img, tuple(i[0]), 10, (0, 0, 255), -1)

    cv2.imshow('src', img)

    cv2.waitKey(0)

When using findCirclesGrid in python it seems that the ordering of the found circles is inverted. My found circles are currently ordered:

• row wise: bottom to top
• column wise: right to left

According to the documentation (at least how I interpret it) I expected it to be the other way around:

• row wise: top to bottom
• column wise: left to right

Several months ago I used this function in c++ and it seemed to behave in the manner as I expected (top to bottom, left to right).

Am I doing something wrong / interpreting the documentation incorrectly or is this a bug that should be reported?

Below a minimal working example which works on the attached bitmap.

edit: image did not get attached. Posted the image here http://imgur.com/a/ROVss

#!/usr/bin/python

http://imgur.com/a/ROVss Minimal working example:

#!/usr/bin/python
import cv2
 
cv2
import numpy as np
 
np
img = cv2.imread('circ_min.bmp', 0)
 
0)
shape = (5, 8)
 
8)
blobPar = cv2.SimpleBlobDetector_Params()
 
cv2.SimpleBlobDetector_Params()
blobPar.minArea = 1e4
 
1e4
blobPar.maxArea = 1e5
 
1e5
blobPar.minDistBetweenBlobs = 200
 
200
blobDet = cv2.SimpleBlobDetector_create(blobPar)
 
cv2.SimpleBlobDetector_create(blobPar)
[found, cc] = cv2.findCirclesGrid(img, 
    shape,
    None,
    flags=cv2.CALIB_CB_SYMMETRIC_GRID+cv2.CALIB_CB_CLUSTERING,
    blobDetector=blobDet)
 
blobDetector=blobDet)

cv2.namedWindow('src', cv2.WINDOW_NORMAL)
 
cv2.WINDOW_NORMAL)
img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
 
cv2.COLOR_GRAY2BGR)
if found:
 
found:
    print(cc)
  for i in cc:
      cv2.circle(img, tuple(i[0]), 10, (0, 0, 255), -1)
      cv2.imshow('src', img)
      cv2.waitKey(0)