def scan_center(xcenter,
ycenter,
start,
stop,
step,
list_coef,
list_power,
output_base,
mat0,
mat_pad,
pad,
axis="x",
ntime=1):
(height, width) = mat0.shape
list_ffact = list_power * list_coef
if axis == "x":
for num in np.arange(start, stop + step, step):
line_img_warped = post.unwarp_image_backward(
mat_pad, xcenter + num + pad, ycenter + pad, list_ffact)
line_img_warped = line_img_warped[pad:pad + height,
pad:pad + width]
name = ("0000" + str(xcenter + num))[-7:]
io.save_image(
output_base + "/xcenter_ntime_" + str(ntime) + "/img_" + name +
".jpg", mat0 + 0.5 * line_img_warped)
else:
for num in range(start, stop + step, step):
line_img_warped = post.unwarp_image_backward(
mat_pad, xcenter + pad, ycenter + num + pad, list_ffact)
line_img_warped = line_img_warped[pad:pad + height,
pad:pad + width]
name = ("0000" + str(ycenter + num))[-7:]
io.save_image(
output_base + "/ycenter_ntime_" + str(ntime) + "/img_" + name +
".jpg", mat0 + 0.5 * line_img_warped)
def test_unwarp_image_backward(self):
x0, y0 = (self.wid // 2, self.hei // 2)
list_fact = [1.0, 3.0 * 10**(-3)]
mat = np.zeros((self.hei, self.wid), dtype=np.float32)
mat[4:-3, 4:-3] = 1.0
mat_warp = post.unwarp_image_backward(mat, x0, y0, list_fact)
vals = np.mean(mat_warp, axis=0)[11:-10]
pos = len(vals) // 2
self.assertTrue(vals[0] < vals[pos] and vals[-1] < vals[pos])
def scan_coef(idx,
start,
stop,
step,
list_coef,
list_power,
output_base,
mat0,
mat_pad,
pad,
ntime=1):
(height, width) = mat0.shape
for num in np.arange(start, stop + step, step):
list_coef_est = np.copy(list_coef)
list_coef_est[idx] = list_coef_est[idx] + num
list_ffact = list_power * list_coef_est
line_img_warped = post.unwarp_image_backward(mat_pad, xcenter + pad,
ycenter + pad, list_ffact)
line_img_warped = line_img_warped[pad:pad + height, pad:pad + width]
name = ("0000" + str(list_coef_est[idx]))[-5:]
io.save_image(
output_base + "/coef_" + str(idx) + "_ntime_" + str(ntime) +
"/img_" + name + ".jpg", mat0 + 0.5 * line_img_warped)
list_fact)
# Calculate the residual of the unwarpped points.
list_hor_data = post.calc_residual_hor(list_uhor_lines, xcenter, ycenter)
list_ver_data = post.calc_residual_ver(list_uver_lines, xcenter, ycenter)
# Save the results for checking
io.save_plot_image(output_base + "/unwarpped_horizontal_lines.png",
list_uhor_lines, height, width)
io.save_plot_image(output_base + "/unwarpped_vertical_lines.png",
list_uver_lines, height, width)
io.save_residual_plot(output_base + "/hor_residual_after_correction.png",
list_hor_data, height, width)
io.save_residual_plot(output_base + "/ver_residual_after_correction.png",
list_ver_data, height, width)
# Correct the image
corrected_mat = post.unwarp_image_backward(mat0, xcenter, ycenter, list_fact)
# Save results.
io.save_image(output_base + "/corrected_image.jpg", corrected_mat)
io.save_image(output_base + "/difference.jpg", corrected_mat - mat0)
# ----------------------------------------
# Load original color image for correction
# ------------------------------------------
# Load coefficients from previous calculation if need to
# (xcenter, ycenter, list_fact) = io.load_metadata_txt(
# output_base + "/coefficients_radial_distortion.txt")
img = io.load_image(file_path, average=False)
img_corrected = np.copy(img)
# Unwarped each color channel of the image
# http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. #============================================================================ # Author: Nghia T. Vo # E-mail: [email protected] #============================================================================ """ Example to show how to apply distortion correction to images of the Hazard Cameras (Hazcams) on the underside of NASA’s Perseverance Mars rover. """ import numpy as np import discorpy.losa.loadersaver as io import discorpy.post.postprocessing as post from PIL import Image # Load color image file_path = "Sol0_1st_color.png" output_base = "figs/" mat = np.asarray(Image.open(file_path), dtype=np.float32) # Import distortion coefficients (xcenter, ycenter, list_fact) = io.load_metadata_txt("figs/coefficients.txt") for i in range(mat.shape[-1]): mat[:, :, i] = post.unwarp_image_backward(mat[:, :, i], xcenter, ycenter, list_fact) io.save_image(output_base + "/Sol0_1st_color_correction.png", mat)
# Save the results for later use.
io.save_metadata_txt(output_base + "/coefficients_radial_distortion.txt",
xcenter, ycenter, list_fact)
print("X-center: {0}. Y-center: {1}".format(xcenter, ycenter))
print("Coefficients: {0}".format(list_fact))
# Apply correction to the lines of points
list_uhor_lines = post.unwarp_line_backward(list_hor_lines, xcenter, ycenter,
list_fact)
list_uver_lines = post.unwarp_line_backward(list_ver_lines, xcenter, ycenter,
list_fact)
# Save the results for checking
io.save_plot_image(output_base + "/unwarpped_horizontal_lines.png", list_uhor_lines,
height, width)
io.save_plot_image(output_base + "/unwarpped_vertical_lines.png", list_uver_lines,
height, width)
# Calculate the residual of the unwarpped points.
list_hor_data = post.calc_residual_hor(list_uhor_lines, xcenter, ycenter)
list_ver_data = post.calc_residual_ver(list_uver_lines, xcenter, ycenter)
# Save the results for checking
io.save_residual_plot(output_base + "/hor_residual_after_correction.png",
list_hor_data, height, width)
io.save_residual_plot(output_base + "/ver_residual_after_correction.png",
list_ver_data, height, width)
# Correct the image
corrected_mat = post.unwarp_image_backward(mat0, xcenter, ycenter, list_fact)
# Save results. Note that the output is 32-bit numpy array. Convert to lower-bit if need to.
io.save_image(output_base + "/corrected_image.tif", corrected_mat)
io.save_image(output_base + "/difference.tif", corrected_mat - mat0)
# Get a good estimation of the forward model
list_ffact = list_coef * list_power
# Transform to the backward model for correction
ref_points = [[i - ycenter, j - xcenter] for i in range(0, height, 50)
for j in range(0, width, 50)]
list_bfact = proc.transform_coef_backward_and_forward(list_ffact,
ref_points=ref_points)
# Load the color image
img = io.load_image(file_path, average=False)
img_corrected = np.copy(img)
# Unwarped each color channel of the image
for i in range(img.shape[-1]):
img_corrected[:, :,
i] = post.unwarp_image_backward(img[:, :, i], xcenter,
ycenter, list_bfact)
# Save the unwarped image.
io.save_image(output_base + "/F_R_hazcam_unwarped.png", img_corrected)
# Correct other images from the same camera
img = io.load_image("C:/data/percy_cam/rock_core1.png", average=False)
img_corrected = np.copy(img)
for i in range(img.shape[-1]):
img_corrected[:, :,
i] = post.unwarp_image_backward(img[:, :, i], xcenter,
ycenter, list_bfact)
io.save_image(output_base + "/rock_core1_unwarped.png", img_corrected)
img = io.load_image("C:/data/percy_cam/rock_core2.png", average=False)
img_corrected = np.copy(img)
import numpy as np
import discorpy.losa.loadersaver as io
import discorpy.post.postprocessing as post
# Load image
mat0 = io.load_image("Sol0_1st_color.png")
output_base = "figs/"
(height, width) = mat0.shape
mat0 = mat0 / np.max(mat0)
# Create line pattern
line_pattern = np.zeros((height, width), dtype=np.float32)
for i in range(50, height - 50, 40):
line_pattern[i - 1:i + 2] = 1.0
# Estimate parameters by visual inspection.
# Coarse estimation
xcenter = width / 2.0 + 110.0
ycenter = height / 2.0 - 20.0
list_pow = np.asarray([1.0, 10**(-4), 10**(-7), 10**(-10), 10**(-13)])
# Fine estimation
list_coef = np.asarray([1.0, 4.0, 5.0, 17.0, 3.0])
list_ffact = list_pow * list_coef
pad = width
mat_pad = np.pad(line_pattern, pad, mode='edge')
mat_cor = post.unwarp_image_backward(mat_pad, xcenter + pad, ycenter + pad,
list_ffact)
mat_cor = mat_cor[pad:pad + height, pad:pad + width]
io.save_image(output_base + "/overlay.jpg", (mat0 + 0.5 * mat_cor))
# xcenter, ycenter, num_coef)
# list_uhor_lines = post.unwarp_line_forward(list_hor_lines2, xcenter, ycenter,
# list_ffact)
# list_uver_lines = post.unwarp_line_forward(list_ver_lines2, xcenter, ycenter,
# list_ffact)
# Check the residual of unwarped lines:
list_hor_data = post.calc_residual_hor(list_uhor_lines, xcenter, ycenter)
list_ver_data = post.calc_residual_ver(list_uver_lines, xcenter, ycenter)
io.save_residual_plot(output_base + "/hor_residual_after_correction.png",
list_hor_data, height, width)
io.save_residual_plot(output_base + "/ver_residual_after_correction.png",
list_ver_data, height, width)
# Unwarp the image
mat_rad_corr = post.unwarp_image_backward(mat0, xcenter, ycenter, list_fact)
# Save results
io.save_image(output_base + "/image_radial_corrected.jpg", mat_rad_corr)
io.save_image(output_base + "/radial_difference.jpg", mat_rad_corr - mat0)
# -------------------------------------
# For perspective distortion correction
# -------------------------------------
# Generate source points and target points to calculate coefficients of a perspective model
source_points, target_points = proc.generate_source_target_perspective_points(
list_uhor_lines,
list_uver_lines,
equal_dist=True,
scale="mean",
optimizing=False)
