params_cam = sensor_params["params_cam"]
rospy.init_node('lane_detector', anonymous=True)
image_parser = IMGParser()
bev_op = BEVTransform(params_cam=params_cam)
curve_learner = CURVEFit(order=1)
rate = rospy.Rate(20)
while not rospy.is_shutdown():
if image_parser.edges is not None:
img_warp = bev_op.warp_bev_img(image_parser.edges)
lane_pts = bev_op.recon_lane_pts(image_parser.edges)
x_pred, y_pred_l, y_pred_r = curve_learner.fit_curve(lane_pts)
curve_learner.write_path_msg(x_pred, y_pred_l, y_pred_r)
curve_learner.pub_path_msg()
xyl, xyr = bev_op.project_lane2img(x_pred, y_pred_l, y_pred_r)
img_warp1 = draw_lane_img(img_warp, xyl[:, 0].astype(np.int32),
xyl[:, 1].astype(np.int32),
xyr[:, 0].astype(np.int32),
xyr[:, 1].astype(np.int32))
pts_learner = CURVEFit(order=3)
lane_detector = LINEDetector()
rate = rospy.Rate(20)
img_wlane = None
while not rospy.is_shutdown():
if lane_detector.cam is not None:
img_wlane = lane_detector.binary_image()
if img_wlane is not None:
# 카메라 이미지를 BEV이미지로 변환
img_bev = bev_trans.warp_bev_img(img_wlane)
# 카메라 이미지에서 차선에 해당하는 포인트들을 추출
lane_pts = bev_trans.recon_lane_pts(img_wlane)
# 추출한 포인트를 기반으로 차선을 예측
x_pred, y_pred_l, y_pred_r = pts_learner.fit_curve(lane_pts)
# 예측한 차선을 Path로 변환하여 메세지의 데이터를 작성
pts_learner.write_path_msg(x_pred, y_pred_l, y_pred_r)
# 예측한 차선을 publish topic: /lane_path
pts_learner.pub_path_msg()
# 예측한 차선 포인트들을 이미지에 넣기 위해서 변환
xyl, xyr = bev_trans.project_lane2img(x_pred, y_pred_l, y_pred_r)
# 예측한 차선 포인트들을 BEV이미지에 넣기
'r') as fp:
sensor_params = json.load(fp)
params_cam = sensor_params["params_cam"]
rospy.init_node('lane_detector', anonymous=True)
image_parser = IMGParser()
bev_op = BEVTransform(params_cam=params_cam)
curve_learner = CURVEFit(order=3)
rate = rospy.Rate(20)
while not rospy.is_shutdown():
img_wlane = image_parser.get_bi_img()
img_warp = bev_op.warp_bev_img(img_wlane)
lane_pts = bev_op.recon_lane_pts(img_wlane)
x_pred, y_pred_l, y_pred_r = curve_learner.fit_curve(lane_pts)
curve_learner.write_path_msg(x_pred, y_pred_l, y_pred_r)
curve_learner.pub_path_msg()
xyl, xyr = bev_op.project_lane2img(x_pred, y_pred_l, y_pred_r)
img_warp1 = draw_lane_img(img_warp, xyl[:, 0].astype(np.int32),
xyl[:, 1].astype(np.int32),
xyr[:, 0].astype(np.int32),
xyr[:, 1].astype(np.int32))
bev_trans = BEVTransform(params_cam=params_cam, cut_size=0.55)
# BEV로 변환된 이미지에서 추출한 포인트를 기반으로 RANSAC을 이용하여 차선을 예측하는 클래스를 선언
pts_learner = CURVEFit(order=3, init_width=0.5)
lane_detector = LINEDetector()
rate = rospy.Rate(20)
while not rospy.is_shutdown():
if lane_detector.img_lane is not None:
img_lane_pts = copy.deepcopy(lane_detector.img_lane)
img_bev = copy.deepcopy(lane_detector.img_lane)
# lane_detector.img_lane[:int(0.6*640), :] = 0
# 카메라 이미지를 BEV이미지로 변환
img_bev = bev_trans.warp_bev_img(img_bev)
# 이진화된 카메라 이미지를 전체에서(위에서) 0.5(전체=1)만큼 자르고 나머지에서 차선에 해당하는 포인트들을 추출
lane_pts = bev_trans.recon_lane_pts(img_lane_pts)
# print(lane_pts[1])
pts_learner.init_setting(lane_pts)
# 추출한 포인트를 기반으로 차선을 예측
x_pred, y_pred_l, y_pred_r = pts_learner.fit_curve(lane_pts)
# 예측한 차선을 Path로 변환하여 메세지의 데이터를 작성
pts_learner.write_path_msg(x_pred, y_pred_l, y_pred_r)
# 예측한 차선을 publish topic: /lane_path
pts_learner.pub_path_msg()
# 예측한 차선 포인트들을 이미지에 넣기 위해서 변환
xyl, xyr = bev_trans.project_lane2img(x_pred, y_pred_l, y_pred_r)
with open(os.path.join(currentPath, 'sensor/sensor_params.json'),
'r') as fp:
sensor_params = json.load(fp)
params_cam = sensor_params["params_cam"]
rospy.init_node('lane_detector', anonymous=True)
image_parser = IMGParser()
bev_op = BEVTransform(params_cam=params_cam)
curve_learner = CURVEFit(order=3)
rate = rospy.Rate(20)
while not rospy.is_shutdown():
if image_parser.img_wlane is not None:
img_warp = bev_op.warp_bev_img(image_parser.img_wlane)
lane_pts = bev_op.recon_lane_pts(image_parser.img_wlane)
x_pred, y_pred_l, y_pred_r = curve_learner.fit_curve(lane_pts)
curve_learner.write_path_msg(x_pred, y_pred_l, y_pred_r)
curve_learner.pub_path_msg()
xyl, xyr = bev_op.project_lane2img(x_pred, y_pred_l, y_pred_r)
img_warp1 = draw_lane_img(img_warp, xyl[:, 0].astype(np.int32),
xyl[:, 1].astype(np.int32),
xyr[:, 0].astype(np.int32),
xyr[:, 1].astype(np.int32))
#cv2.imshow("lane_detector", img_warp1)