def main():
# initial threshold
th_light = 75
cv2.namedWindow('Result')
cv2.createTrackbar('Luminance:', 'Result', th_light, 99, callback_ths)
for _ in range(20000):
# record time
t_s = time.time()
# image parsing
img_cam = get_img(UDP_cam, params_cam)
img_binary = binary_pipeline(img_cam, th_light, (80, 95)) #lane
img_binary_warp = warp_image(img_binary, source_points)
img_morph = morphologic_process(img_binary_warp, 5, 1, 1)
t_cost = time.time()-t_s
fit_check, left_fit, right_fit = fit_track_lanes(img_morph, poly_order=poly_order,
nwindows=30,
margin=30,
minpix=20)
img_morph = cv2.cvtColor(img_morph*255, cv2.COLOR_GRAY2BGR)
if fit_check:
ploty = np.linspace(0, params_cam["HEIGHT"]-1, params_cam["HEIGHT"])
left_fitx = get_poly_val(ploty, left_fit)
right_fitx = get_poly_val(ploty, right_fit)
left_fitx, left_fity = crop_points(left_fitx.astype(np.int32), ploty.astype(np.int32), params_cam["WIDTH"], params_cam["HEIGHT"])
right_fitx, right_fity = crop_points(right_fitx.astype(np.int32), ploty.astype(np.int32), params_cam["WIDTH"], params_cam["HEIGHT"])
img_line = draw_lane_img(img_morph, left_fitx, left_fity, right_fitx, right_fity)
else:
img_line = np.copy(img_morph)
th_light = visualize_images([cv2.polylines(img_cam,[source_points.astype(np.int32)],True,(255,0,0), 5),
cv2.cvtColor(img_binary*255, cv2.COLOR_GRAY2BGR),
img_morph,
img_line],
t_cost, params_cam, img_name=['src', 'binary', 'morph', 'lane'])
UDP_cam.close()
def binary_pipeline_test():
# initial threshold
th_light = 75
cv2.namedWindow('Result')
cv2.createTrackbar('Luminance:', 'Result', th_light, 99, callback_ths)
for _ in range(20000):
# record time
t_s = time.time()
# load the image
# img_cam = cv2.imread('./image_sample/image_straight.jpg')
img_cam = cv2.imread(
'/home/junho/Autonomous/HENVEN_PathPlanning/Lane_Detection/Lane_image/lane_1.jpg'
)
img_cam = cv2.resize(img_cam,
dsize=(480, 320),
interpolation=cv2.INTER_LINEAR)
img_binary = binary_pipeline(img_cam, th_light, (80, 95)) #lane
img_binary_warp = warp_image(img_binary, source_points)
img_morph = morphologic_process(img_binary_warp, 5, 1, 1)
t_cost = time.time() - t_s
img_final = cv2.cvtColor(img_morph * 255, cv2.COLOR_GRAY2BGR)
th_light = visualize_images([
cv2.polylines(img_cam, [source_points.astype(np.int32)], True,
(255, 0, 0), 5),
cv2.cvtColor(img_binary_warp * 255, cv2.COLOR_GRAY2BGR), img_final
],
t_cost,
params_cam,
img_name=['src', 'warp', 'morph'])
def lane_detection(image):
#initial threshold
th_light = 95
cv2.namedWindow('Result')
#cv2.createTrackbar('Luminance:', 'Result', th_light, 99, callback_ths)
# lane function order
poly_order = 1
for _ in range(20000):
# record time
t_s = time.time()
#resize image
img = cv2.resize(image, dsize=(480, 320))
img_binary = binary_pipeline(img, th_light, (80, 95))
img_binary_warp = warp_image(img_binary, source_points)
img_morph = morphologic_process(img_binary_warp, 5, 1, 1)
#t_cost = time.time()-t_s
fit_check, left_fit, right_fit = fit_track_lanes(img_morph,
poly_order=poly_order,
nwindows=30,
margin=30,
minpix=20)
img_morph = cv2.cvtColor(img_morph * 255, cv2.COLOR_GRAY2BGR)
if fit_check == str('two_lanes'):
ploty = np.linspace(0, params_img["HEIGHT"] - 1,
params_img["HEIGHT"])
left_fitx = get_poly_val(ploty, left_fit)
right_fitx = get_poly_val(ploty, right_fit)
left_fitx, left_fity = crop_points(left_fitx.astype(np.int32),
ploty.astype(np.int32),
params_img["WIDTH"],
params_img["HEIGHT"])
right_fitx, right_fity = crop_points(right_fitx.astype(np.int32),
ploty.astype(np.int32),
params_img["WIDTH"],
params_img["HEIGHT"])
img_line = draw_lane_img(img_morph, left_fitx, left_fity,
right_fitx, right_fity)
tracking_x, tracking_y = img_lane_tracking_point(
left_fitx, left_fity, right_fitx, right_fity)
base_x = img_line.shape[1] / 2
base_y = img_line.shape[0]
img_tracking_point_lane = draw_tracking_point_lane_img(
img_line, tracking_x, tracking_y, base_x, base_y)
angle = calculate_angle(tracking_x, tracking_y, base_x, base_y)
elif fit_check == str('central_lane'):
ploty = np.linspace(0, params_img["HEIGHT"] - 1,
params_img["HEIGHT"])
central_fit = left_fit
central_fitx = get_poly_val(ploty, central_fit)
central_fitx, central_fity = crop_points(
central_fitx.astype(np.int32), ploty.astype(np.int32),
params_img["WIDTH"], params_img["HEIGHT"])
img_line = draw_central_lane_img(img_morph, central_fitx,
central_fity)
base_x = img_line.shape[1] / 2
base_y = img_line.shape[0]
angle, tracking_x, tracking_y = img_lane_tracking_point_single(
central_fitx, central_fity, base_x)
img_tracking_point_lane = draw_tracking_point_lane_img(
img_line, tracking_x, tracking_y, base_x, base_y)
elif fit_check == str('left_lane'):
ploty = np.linspace(0, params_img["HEIGHT"] - 1,
params_img["HEIGHT"])
left_fitx = get_poly_val(ploty, left_fit)
left_fitx, left_fity = crop_points(left_fitx.astype(np.int32),
ploty.astype(np.int32),
params_img["WIDTH"],
params_img["HEIGHT"])
img_line = draw_left_lane_img(img_morph, left_fitx, left_fity)
base_x = img_line.shape[1] / 2
base_y = img_line.shape[0]
angle, tracking_x, tracking_y = img_lane_tracking_point_single(
left_fitx, left_fity, base_x)
img_tracking_point_lane = draw_tracking_point_lane_img(
img_line, tracking_x, tracking_y, base_x, base_y)
elif fit_check == str('right_lane'):
ploty = np.linspace(0, params_img["HEIGHT"] - 1,
params_img["HEIGHT"])
right_fitx = get_poly_val(ploty, right_fit)
right_fitx, right_fity = crop_points(right_fitx.astype(np.int32),
ploty.astype(np.int32),
params_img["WIDTH"],
params_img["HEIGHT"])
img_line = draw_right_lane_img(img_morph, right_fitx, right_fity)
base_x = img_line.shape[1] / 2
base_y = img_line.shape[0]
angle, tracking_x, tracking_y = img_lane_tracking_point_single(
right_fitx, right_fity, base_x)
img_tracking_point_lane = draw_tracking_point_lane_img(
img_line, tracking_x, tracking_y, base_x, base_y)
else:
img_line = np.copy(img_morph)
tracking_x = img_line.shape[1] / 2
tracking_y = img_line.shape[0] / 2
base_x = img_line.shape[1] / 2
base_y = img_line.shape[0]
img_tracking_point_lane = draw_tracking_point_lane_img(
img_line, tracking_x, tracking_y, base_x, base_y)
angle = calculate_angle(tracking_x, tracking_y, base_x, base_y)
print(fit_check)
t_cost = time.time() - t_s
th_light = visualize_images_angle([
cv2.polylines(img, [source_points.astype(np.int32)], True,
(255, 0, 0), 5), img_tracking_point_lane
],
t_cost,
params_img,
angle,
img_name=['src', 'lane'])
return img_tracking_point_lane, angle