ArUco Marker Detection

Detect and track ArUco markers for augmented reality.

View Source Code

Overview

ArUco markers are binary square fiducial markers used in computer vision for camera pose estimation and augmented reality applications.

Key Techniques:

  • ArUco dictionary generation
  • Marker detection
  • Pose estimation
  • 3D axis drawing

ArUco Dictionaries

Dictionary Markers Best For
DICT_4X4_50 50 Small markers, close range
DICT_5X5_100 100 Medium markers
DICT_6X6_250 250 Large markers, far range
DICT_ARUCO_ORIGINAL 1024 Original ArUco library

Key OpenCV Functions 

# Get dictionary
aruco_dict = cv2.aruco.getPredefinedDictionary(cv2.aruco.DICT_4X4_50)
parameters = cv2.aruco.DetectorParameters()
detector = cv2.aruco.ArucoDetector(aruco_dict, parameters)

# Detect markers
corners, ids, rejected = detector.detectMarkers(gray)

# Draw detected markers
cv2.aruco.drawDetectedMarkers(frame, corners, ids)

# Generate a marker image
marker = cv2.aruco.generateImageMarker(aruco_dict, marker_id=0, sidePixels=200)

Pose Estimation 

# Camera matrix (should be from calibration)
focal_length = frame.shape[1]
center = (frame.shape[1] / 2, frame.shape[0] / 2)
camera_matrix = np.array([
    [focal_length, 0, center[0]],
    [0, focal_length, center[1]],
    [0, 0, 1]
], dtype=np.float32)
dist_coeffs = np.zeros((4, 1))

# Marker size in meters
marker_size = 0.05

# Define 3D points
obj_points = np.array([
    [-marker_size/2, marker_size/2, 0],
    [marker_size/2, marker_size/2, 0],
    [marker_size/2, -marker_size/2, 0],
    [-marker_size/2, -marker_size/2, 0]
], dtype=np.float32)

# Solve PnP for each marker
for corner in corners:
    success, rvec, tvec = cv2.solvePnP(
        obj_points, corner[0], camera_matrix, dist_coeffs
    )
    if success:
        cv2.drawFrameAxes(frame, camera_matrix, dist_coeffs, rvec, tvec, 0.03)

Generate Marker Board 

def generate_board(rows=2, cols=3, marker_size=100, margin=20):
    board_width = cols * marker_size + (cols + 1) * margin
    board_height = rows * marker_size + (rows + 1) * margin
    board = np.ones((board_height, board_width), dtype=np.uint8) * 255

    marker_id = 0
    for r in range(rows):
        for c in range(cols):
            marker = cv2.aruco.generateImageMarker(aruco_dict, marker_id, marker_size)
            x = margin + c * (marker_size + margin)
            y = margin + r * (marker_size + margin)
            board[y:y+marker_size, x:x+marker_size] = marker
            marker_id += 1

    return board

Applications

  • Augmented Reality: Overlay 3D objects on markers
  • Robot Navigation: Localization using markers
  • Camera Calibration: Known-size markers
  • Object Tracking: Track marker position and orientation

Controls

Key Action
1-4 Change dictionary
p Toggle pose estimation
g Generate and save marker
s Save screenshot
q Quit

Running the Application 

python curriculum/applications/11_aruco_detection.py

Note: Requires opencv-contrib-python for ArUco module.

Official Documentation