0
$\begingroup$

Rosanswers logo

Hello

I can currently identify an aruco marker with my webcam. I would like to calculate the distance "D" between the webcam and the Aruco marker. Is it possible to find the angle "a" between the blue axis of the Aruco marker and the center of the camera image?

note: 3m distance is an example, I can get it through a rangefinder that measures the distance from the center of the camera to the center of the Arcuo marker. The camera is fixed at the center of a differential robot and the robot can align with the center of the Aruco marker. I would like to calculate the distance "D" to position(approximately) the robot in the center of the Aruco marker.

image description

image description

i'm using kyle-bersani code (https://github.com/kyle-bersani/opencv-examples):

import numpy 
import cv2
import cv2.aruco as aruco
import os
import pickle

# Check for camera calibration data
if not os.path.exists('./calibration.pckl'):
    print("You need to calibrate the camera you'll be using. See calibration project directory for details.")
    exit()
else:
    f = open('calibration.pckl', 'rb')
    (cameraMatrix, distCoeffs, _, _) = pickle.load(f)
    f.close()
    if cameraMatrix is None or distCoeffs is None:
        print("Calibration issue. Remove ./calibration.pckl and recalibrate your camera with CalibrateCamera.py.")
        exit()

# Constant parameters used in Aruco methods
ARUCO_PARAMETERS = aruco.DetectorParameters_create()
#ARUCO_DICT = aruco.Dictionary_get(aruco.DICT_6X6_1000) original
ARUCO_DICT = aruco.Dictionary_get(aruco.DICT_5X5_1000)

# Create grid board object we're using in our stream
board = aruco.GridBoard_create(
        markersX=2,
        markersY=2,
        markerLength=0.09,
        markerSeparation=0.01,
        dictionary=ARUCO_DICT)

# Create vectors we'll be using for rotations and translations for postures
rvecs, tvecs = None, None

cam = cv2.VideoCapture(0)

while(cam.isOpened()):
    # Capturing each frame of our video stream
    ret, QueryImg = cam.read()
    if ret == True:
        # grayscale image
        gray = cv2.cvtColor(QueryImg, cv2.COLOR_BGR2GRAY)

        # Detect Aruco markers
        corners, ids, rejectedImgPoints = aruco.detectMarkers(gray, ARUCO_DICT, parameters=ARUCO_PARAMETERS)

        # Refine detected markers
        # Eliminates markers not part of our board, adds missing markers to the board
        corners, ids, rejectedImgPoints, recoveredIds = aruco.refineDetectedMarkers(
                image = gray,
                board = board,
                detectedCorners = corners,
                detectedIds = ids,
                rejectedCorners = rejectedImgPoints,
                cameraMatrix = cameraMatrix,
                distCoeffs = distCoeffs)   

        
        QueryImg = aruco.drawDetectedMarkers(QueryImg, corners, borderColor=(0, 0, 255))

    if ids is not None:
        try:
            rvec, tvec, _objPoints = aruco.estimatePoseSingleMarkers(corners, 10.5, cameraMatrix, distCoeffs)                                           
            QueryImg = aruco.drawAxis(QueryImg, cameraMatrix, distCoeffs, rvec, tvec, 5)           
        except:
            print("Deu merda segue o baile")
    
        
        cv2.imshow('QueryImage', QueryImg)

    # Exit at the end of the video on the 'q' keypress
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

cv2.destroyAllWindows()

thanks


Originally posted by mateusguilherme on ROS Answers with karma: 125 on 2020-01-14

Post score: 0


Original comments

Comment by mateusguilherme on 2020-01-14:
I think I need to find the angle between the normal vector (blue axis) and the center of the camera. Then I could calculate the distance "D" through trigonometry: D = 3 * sen(a).

note: maybe there is another way to calculate the distance "D" that I don't know

Comment by gvdhoorn on 2020-01-14:
Isn't D just the translation along the negative Y-axis of the transform between Aruco Marker and the pose of the web cam?

Comment by stevemacenski on 2020-01-14:
Doesn't the fiducial library just give you the pose of the marker?

$\endgroup$

1 Answer 1

0
$\begingroup$

Rosanswers logo

for those who are facing a similar problem, look for the Rodrigues function. From it it is possible to obtain roll pitch and yaw in relation to the marker


Originally posted by mateusguilherme with karma: 125 on 2020-02-23

This answer was ACCEPTED on the original site

Post score: 0


Original comments

Comment by siddharthcb on 2021-05-13:
were you able to identify how to get D?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.