Stereo camera calibration via openCV chessboard

Question

I am trying to calibrate a stereo camera using openCV, but the function findChessboardCorners is always returning False, I believe it is in function of high image distortion. Any hint of preprocessing to make the function acerctive?

Code example:

# Encontre as Bordas do tabuleiro
ret, corners = cv2.findChessboardCorners(img, (5, 4),cv2.CALIB_CB_ASYMMETRIC_GRID)

Example of the images:

Answer 1

With ret, corners = cv2.findChessboardCorners(img, (5, 4),cv2.CALIB_CB_ASYMMETRIC_GRID), you are using the pattern matrix (5, 4).

But if you look at the printed chessboard, the pattern sheet is (7, 9) and 20 mm x 20 mm.

Then using the official Calibration tutorial, the following code is created:

Code

import numpy as np
import cv2 as cv
import glob

# Dimensões do Tabuleiro de Xadrez
cbcol = 7
cbrow = 9
cbw = 20

# Critério
criteria = (cv.TERM_CRITERIA_EPS + cv.TERM_CRITERIA_MAX_ITER, cbw, 0.001)


# preparar os pontos do objeto, como (0,0,0), (1,0,0), (2,0,0) ....,(6,5,0)
objp = np.zeros((cbrow * cbcol, 3), np.float32)
objp[:, :2] = np.mgrid[0:cbcol, 0:cbrow].T.reshape(-1, 2)
# Vetores para armazenar os pontos de objeto e pontos de imagem de todas as imagens.
objpoints = [] # ponto 3d no espaço do mundo real
imgpoints = [] # ponto 2d no plano da imagem.
images = glob.glob('C:\\Users\\Desktop\\teste\\calib\\*.jpg')
i=0
for fname in images:
    print(fname)
    img = cv.imread(fname)
    gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY)
    # Encontra os cantos do tabuleiro de xadrez
    ret, corners = cv.findChessboardCorners(gray, (cbcol,cbrow), None)
    # Se encontrado, adiciona os pontos de objeto e pontos de imagem (após refiná-los)
    print(ret)
    if ret == True:
        objpoints.append(objp)
        corners2 = cv.cornerSubPix(gray,corners, (11,11), (-1,-1), criteria)
        imgpoints.append(corners)
        # Desenha e mostra os cantos
        cv.drawChessboardCorners(img, (cbcol, cbrow), corners2, ret)
        cv.imwrite('C:\\Users\\Desktop\\teste\\calib\\resultado\\' + str(i) + '.jpg', img)
        cv.imshow('img', img)
        cv.waitKey(500)
        i += 1
cv.destroyAllWindows()

Results

Remembering that the official tutorial asks for at least 10 photos to perform a reliable calibration. Or this problem may occur.

Calibrate Camera

After finding the points, calibration can be performed:

ret, mtx, dist, rvecs, tvecs = cv.calibrateCamera(objpoints, imgpoints, gray.shape[::-1], None, None)

Create new camera matrix

h,  w = img.shape[:2]
newcameramtx, roi = cv.getOptimalNewCameraMatrix(mtx, dist, (w,h), 1, (w,h))

Undistorted image

There are two methods to accomplish this, the first uses the Opencv function undistort() and the second the Opencv function remap()

undistort()

dst = cv.undistort(img, mtx, dist, None, newcameramtx)
# crop the image
x, y, w, h = roi
dst = dst[y:y+h, x:x+w]
cv.imwrite('calibresult.png', dst)

remap()

mapx, mapy = cv.initUndistortRectifyMap(mtx, dist, None, newcameramtx, (w,h), 5)
dst = cv.remap(img, mapx, mapy, cv.INTER_LINEAR)
# crop the image
x, y, w, h = roi
dst = dst[y:y+h, x:x+w]
cv.imwrite('calibresult.png', dst)