"rtree-c/test/expected/for/reduction/r0000011111.c" did not exist on "3938eff9a94739fcdfa0c78286f3aa18c054e0e4"
Newer
Older
import time
import cv2
import numpy as np
import matplotlib.pyplot as plt
from skimage import exposure
from skimage.filters import gaussian
from skimage.feature import canny
from skimage.graph import route_through_array
Load an image in either gray or color mode (then convert color to gray).
"""
if type == 'gray':
img = cv2.imread(path, cv2.IMREAD_GRAYSCALE)
if img is None:
raise FileNotFoundError(f"Could not read {path}")
elif type == 'color':
img = cv2.imread(path, cv2.IMREAD_COLOR)
if img is None:
raise FileNotFoundError(f"Could not read {path}")
else:
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
else:
raise ValueError("type must be 'gray' or 'color'")
return img
def downscale(img, points, scale_percent):
"""
Downsample `img` to `scale_percent` size and scale the given points accordingly.
Returns (downsampled_img, (scaled_seed, scaled_target)).
"""
if scale_percent == 100:
return img, (tuple(points[0]), tuple(points[1]))
else:
width = int(img.shape[1] * scale_percent / 100)
height = int(img.shape[0] * scale_percent / 100)
new_dimensions = (width, height)
downsampled_img = cv2.resize(img, new_dimensions, interpolation=cv2.INTER_AREA)
# Scaling factors
scale_x = width / img.shape[1]
scale_y = height / img.shape[0]
seed_xy = tuple(points[0])
target_xy = tuple(points[1])
scaled_seed_xy = (int(seed_xy[0] * scale_x), int(seed_xy[1] * scale_y))
scaled_target_xy = (int(target_xy[0] * scale_x), int(target_xy[1] * scale_y))
return downsampled_img, (scaled_seed_xy, scaled_target_xy)
def compute_cost(image, sigma=3.0, epsilon=1e-5):
"""
Smooth the image, run Canny edge detection, then invert the edge map into a cost image.
"""
# Apply histogram equalization
image_contrasted = exposure.equalize_adapthist(image, clip_limit=0.01)
# Apply smoothing
smoothed_img = gaussian(image_contrasted, sigma=sigma)
# Apply Canny edge detection
# Create cost image
cost_img = 1.0 / (canny_img + epsilon) # Invert edges: higher cost where edges are stronger
def backtrack_pixels_on_image(img_color, path_coords, bgr_color=(0, 0, 255)):
"""
Color the path on the (already converted BGR) image in the specified color.
`path_coords` should be a list of (row, col) or (y, x).
"""
for (row, col) in path_coords:
img_color[row, col] = bgr_color
return img_color
def export_path(path_coords, path_name):
"""
Export the path to a np array.
"""
np.save(path_name, path_coords)
return None
#### Main Script ####
def main():
# Define input parameters
image_type = 'gray' # 'gray' or 'color'
downscale_factor = 100 # % of original size
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
# Load image
image = load_image(image_path, image_type)
# Load seed and target points
points = np.int0(np.round(np.load(points_path))) # shape: (2, 2), i.e. [[x_seed, y_seed], [x_target, y_target]]
# Downscale image and points
scaled_image, scaled_points = downscale(image, points, downscale_factor)
seed, target = scaled_points # Each is (x, y)
# Convert to row,col for scikit-image (which uses (row, col) = (y, x))
seed_rc = (seed[1], seed[0])
target_rc = (target[1], target[0])
# Compute cost image
cost_image, canny_img = compute_cost(scaled_image)
# Find path using route_through_array
# route_through_array expects: route_through_array(image, start, end, fully_connected=True/False)
start_time = time.time()
path_rc, cost = route_through_array(
cost_image,
start=seed_rc,
end=target_rc,
fully_connected=True
)
end_time = time.time()
print(f"Elapsed time for pathfinding: {end_time - start_time:.3f} seconds")
# Convert single-channel image to BGR for coloring
color_img = cv2.cvtColor(scaled_image, cv2.COLOR_GRAY2BGR)
# Draw path. `path_rc` is a list of (row, col).
# If you want to mark it in red, do (0,0,255) because OpenCV uses BGR format.
color_img = backtrack_pixels_on_image(color_img, path_rc, bgr_color=(0, 0, 255))
# Export path
export_path(path_rc, 'agamodonPath.npy')
# Display results
plt.figure(figsize=(20, 8))
plt.subplot(1, 2, 1)
plt.title("Cost image")
plt.imshow(cost_image, cmap='gray')
plt.subplot(1, 2, 2)
plt.title("Path from Seed to Target")
# Convert BGR->RGB for pyplot
plt.imshow(color_img[..., ::-1])
plt.show()
if __name__ == "__main__":