我正在使用英特尔实感 D415 摄像头来测量到物体的距离并获得相应的 3D 坐标。我得到的坐标是准确的,但是当我尝试将它们转换为另一个坐标系时出现问题。
我想做的是:
第 1 步和第 2 步工作正常。我的问题在第 3 步。我使用 rs2_deproject_pixel_to_point 来获取正确的坐标。我的理解是,我可以校准第二个外在矩阵之类的东西,并使用以下方法转换新系统中的点:
P_world = R * P_Camera + t = Rt * P_Camera
Rt 为 3x4 矩阵,P_Camera 为 4x1 向量
为了获得矩阵Rt,我打印了一个棋盘并将其放置在我的世界坐标中的已知位置。我选择桌子的一角作为原点。现在我正在 RGB 帧中搜索棋盘并保存深度帧中的 3D 相机坐标。在这一步之后,我试图用来自 4 个具有不同 X、Y、Z 值和已知角位置的点的 12 个线性方程求解矩阵。到目前为止一切顺利,但我得到的结果远非正确,我不知道出了什么问题。
是方法本身的错误还是只是我的实现导致了错误的值?有什么方法可以轻松地转换另一个系统中的坐标吗?
以下是我用来查找棋盘和获取 3D 位置的代码:
def convert_depth_to_phys_coord_using_realsense(x, y, depth, cameraInfo):
_intrinsics = rs.intrinsics()
_intrinsics.width = cameraInfo.width
_intrinsics.height = cameraInfo.height
_intrinsics.ppx = cameraInfo.ppx
_intrinsics.ppy = cameraInfo.ppy
_intrinsics.fx = cameraInfo.fx
_intrinsics.fy = cameraInfo.fy
# _intrinsics.model = cameraInfo.distortion_model
_intrinsics.model = rs.distortion.none
_intrinsics.coeffs = [i for i in cameraInfo.coeffs]
result = rs.rs2_deproject_pixel_to_point(_intrinsics, [x, y], depth)
# result[0]: right, result[1]: down, result[2]: forward
return result[0], -result[1], -result[2]
...
# Setup the pipeline
pipe = rs.pipeline()
cfg = rs.config()
cfg.enable_stream(rs.stream.depth, 1280, 720, rs.format.z16, 30)
cfg.enable_stream(rs.stream.color, 1280, 720, rs.format.bgr8, 30)
profile = pipe.start(cfg)
# Setup the 'High Accuracy'-mode
depth_sensor = profile.get_device().first_depth_sensor()
preset_range = depth_sensor.get_option_range(rs.option.visual_preset)
for i in range(int(preset_range.max)):
visulpreset = depth_sensor.get_option_value_description(rs.option.visual_preset,i)
print('%02d: %s'%(i,visulpreset))
if visulpreset == "High Accuracy":
depth_sensor.set_option(rs.option.visual_preset, i)
# enable higher laser-power for better detection
depth_sensor.set_option(rs.option.laser_power, 180)
# lower the depth unit for better accuracy and shorter distance covered
depth_sensor.set_option(rs.option.depth_units, 0.0005)
# Skip first frames for auto-exposure to adjust
for x in range(5):
pipe.wait_for_frames()
while(1):
# Stores next frameset
frameset = pipe.wait_for_frames()
color_frame = frameset.get_color_frame()
depth_frame = frameset.get_depth_frame()
if color_frame:
color = np.asanyarray(color_frame.get_data())
depth = np.asanyarray(depth_frame.get_data())
# Create alignment primitive with color as its target stream:
align = rs.align(rs.stream.color)
frameset = align.process(frameset)
# Update color and depth frames:
colorizer = rs.colorizer()
aligned_depth_frame = frameset.get_depth_frame()
colorized_depth = np.asanyarray(colorizer.colorize(aligned_depth_frame).get_data())
# termination criteria
criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 30, 0.001)
# processing
# resize the image to match the size of the depth-frame
gray = cv2.cvtColor(color, cv2.COLOR_BGR2GRAY)
resize = cv2.resize(gray,(1280,720))
# Find the chessboard corners
ret, corners = cv2.findChessboardCorners(resize, (rows, columns), None)
#corners2 = cv2.cornerSubPix(gray, corners, (11, 11), (-1, -1), criteria)
if ret:
detected = cv2.drawChessboardCorners(resize, (7,9), corners, ret)
for corner in corners:
x,y = corner.ravel()
dist = aligned_depth_frame.get_distance(x, y)
camera_info = aligned_depth_frame.profile.as_video_stream_profile().intrinsics
x_w, y_w, z_w = convert_depth_to_phys_coord_using_realsense(x, y, dist, camera_info)