def colorThresholding(img_name, mtx, dist, M, thresholds=(60, 255)):
# preprocess image
warped = warpImage(img_name, mtx, dist, M, hls=1)
# thresholding
img_bin = np.zeros_like(warped)
img_bin[(warped >= thresholds[0]) & (warped <= thresholds[1])] = 1
return img_bin
def colorThresholding(img_name, thresholds=(60, 255)):
mtx, dist, M, offset = loadCalibrationVariables()
# preprocess image
warped = warpImage(img_name, mtx, dist, M, hls=1)
# thresholding
img_bin = np.zeros_like(warped)
img_bin[(warped >= thresholds[0]) & (warped <= thresholds[1])] = 1
return img_bin
def gradThresholding(img_name, mtx, dist, M, thresholds=(20, 100)):
# preprocess image
warped = warpImage(img_name, mtx, dist, M)
# compute x-wise gradient and take the absolute value
sobel_x = cv2.Sobel(warped, cv2.CV_64F, 1, 0)
sobel_x_abs = np.absolute(sobel_x)
# scale the absolute values
scaled = np.uint8(255 * sobel_x_abs / np.max(sobel_x_abs))
# apply thresholding
sobel_bin = np.zeros_like(scaled)
sobel_bin[(scaled >= thresholds[0]) & (scaled <= thresholds[1])] = 1
return sobel_bin
white_3 = cv2.inRange(warped, (200, 200, 200), (255, 255, 255))
bit_layer = img_bin | yellow | white | white_2 | white_3
return bit_layer - 1
if __name__ == "__main__":
#=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
# COLOR THRESHOLDING
#=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
mtx, dist, M, offset = loadCalibrationVariables()
# load, extract S channel, and warp example image
img_name = "test_images/test2.jpg"
img_warp = warpImage(img_name, mtx, dist, M, hls=1)
# initialise binary image
img_bin = np.zeros_like(img_warp)
# select thresolds
S_thresholds = (60, 255)
img_bin[(img_warp >= S_thresholds[0]) & (img_warp <= S_thresholds[1])] = 1
# plot images
sns.set_style("white")
fig = plt.figure(1, figsize=(9, 3))
fig.clf()
ax1 = fig.add_subplot(121)
ax1.imshow(img_warp, cmap='gray')
ax1.set_xlabel("$x$")