Spaces:
Build error
Build error
File size: 32,764 Bytes
ec6ad2f bd1f76d ec6ad2f c59b15b 6d5c490 c59b15b ec6ad2f bd1f76d f0c23ec ec6ad2f f0c23ec ec6ad2f f0c23ec ec6ad2f f0c23ec ec6ad2f f0c23ec ec6ad2f f0c23ec ec6ad2f b269113 bd1f76d ec6ad2f bd1f76d ec6ad2f b269113 ec6ad2f bd1f76d e05ccab 76634d5 c59b15b bd1f76d e05ccab bd1f76d e05ccab b269113 e05ccab b269113 e05ccab bd1f76d e05ccab bd1f76d ec6ad2f bd1f76d 6d5c490 bd1f76d 6d5c490 79125e3 bd1f76d 6d5c490 bd1f76d 6d5c490 bd1f76d 6d5c490 bd1f76d 79125e3 bd1f76d ec6ad2f bd1f76d 74018ab bd1f76d 74018ab bd1f76d 74018ab bd1f76d 74018ab bd1f76d 74018ab bd1f76d 74018ab bd1f76d 74018ab bd1f76d 74018ab bd1f76d 6d5c490 bd1f76d 6d5c490 2353a2a 6d5c490 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 |
from typing import List, Tuple
from pathlib import Path
from .config import Config
import numpy as np
import cv2
from skimage.morphology import skeletonize, remove_small_objects
from skimage.measure import label
from skimage import measure
from tqdm import tqdm
from PIL import Image
import numpy as np
from sklearn.cluster import KMeans
import math
class ImageProcessor:
"""Handles image preprocessing operations."""
def __init__(self, config: Config = None):
self.config = config or Config()
self.index = 0
def get_output_path(self, output_folder, file_name):
self.index += 1
return f'{output_folder}/{self.index:02d}_{file_name}'
def mask_text_regions(self, input_path, bboxes: List[List[int]], output_filename: str = "1_text_removed.jpg", color: Tuple[int, int, int] = (0, 0, 0)) -> str:
"""Mask text regions in the image to reduce panel extraction noise."""
image = cv2.imread(input_path)
if image is None:
raise FileNotFoundError(f"Could not load image: {input_path}")
for bbox in bboxes:
x1, y1, x2, y2 = bbox
cv2.rectangle(image, (x1, y1), (x2, y2), color, thickness=-1)
output_path = f'{self.config.output_folder}/{output_filename}'
cv2.imwrite(output_path, image)
return str(output_path)
def preprocess_image(self, processed_image_path) -> Tuple[str, str, str]:
"""Preprocess image for panel extraction."""
image = cv2.imread(processed_image_path)
if image is None:
raise FileNotFoundError(f"Could not load image: {processed_image_path}")
# Convert to grayscale and binary
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# Apply Gaussian blur to reduce noise
blurred = cv2.GaussianBlur(gray, (3, 3), 0)
# Canny edge detection
edges = cv2.Canny(blurred, threshold1=50, threshold2=150, apertureSize=3)
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 5))
dilated = cv2.dilate(edges, kernel, iterations=2)
# Save intermediate results
gray_path = self.get_output_path(self.config.output_folder, "gray.jpg")
binary_path = self.get_output_path(self.config.output_folder, "binary.jpg")
dilated_path = self.get_output_path(self.config.output_folder, "dilated.jpg")
cv2.imwrite(str(gray_path), gray)
cv2.imwrite(str(binary_path), edges)
cv2.imwrite(str(dilated_path), dilated)
return str(gray_path), str(binary_path), str(dilated_path)
def invert_if_black_dominates(self, binary):
# Threshold to binary image
_, binary = cv2.threshold(binary, 127, 255, cv2.THRESH_BINARY)
# Count black and white pixels
black_pixels = np.sum(binary == 0)
white_pixels = np.sum(binary == 255)
# If black dominates, invert
if black_pixels > white_pixels:
print("🔄 Inverting image because black > white")
inverted = cv2.bitwise_not(binary)
else:
print("✅ No inversion needed, white >= black")
inverted = binary
# Save result
return inverted, black_pixels > white_pixels
def group_colors(self, processed_image_path, num_clusters: int = 5, file_name="group_colors.jpg", output_folder=None) -> Image.Image:
"""
Groups similar colors in an image using KMeans clustering.
Args:
processed_image_path (str): Path to the image to be color-grouped.
num_clusters (int): Number of color clusters to form.
file_name (str): Name of the output image file.
output_folder (str): Optional output directory.
Returns:
str: Path to the saved grouped-color image.
"""
output_folder = output_folder or self.config.output_folder
# Load image
image = Image.open(processed_image_path).convert("RGB")
np_image = np.array(image)
h, w = np_image.shape[:2]
pixels = np_image.reshape(-1, 3)
# Run KMeans
kmeans = KMeans(n_clusters=num_clusters, random_state=42, n_init='auto')
labels = kmeans.fit_predict(pixels)
centers = kmeans.cluster_centers_.astype(np.uint8)
# Replace pixels with their cluster center color
clustered_pixels = centers[labels].reshape(h, w, 3)
# Save using OpenCV (convert RGB to BGR)
output_path = self.get_output_path(output_folder, file_name)
clustered_bgr = clustered_pixels[:, :, ::-1]
cv2.imwrite(output_path, clustered_bgr)
return str(output_path)
def thin_image_borders(self, processed_image_path: str, file_name="thin_border.jpg", output_folder=None) -> str:
"""
Clean dilated image by thinning thick borders and removing hanging clusters.
"""
output_folder = output_folder or self.config.output_folder
# Load image
img = cv2.imread(processed_image_path)
# Convert to grayscale and binary
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# _, binary = cv2.threshold(gray, 200, 255, cv2.THRESH_BINARY_INV)
# Apply Gaussian blur to reduce noise
blurred = cv2.GaussianBlur(gray, (3, 3), 0)
# Canny edge detection
edges = cv2.Canny(blurred, threshold1=50, threshold2=150, apertureSize=3)
# Skeletonize
skeleton = skeletonize(edges).astype(np.uint8)
# Remove small hanging clusters
labeled = label(skeleton, connectivity=2)
cleaned = remove_small_objects(labeled, min_size=150) # Adjust min_size for more/less pruning
# Convert back to 0–255 uint8 image
final = (cleaned > 0).astype(np.uint8) * 255
# Invert back if needed
result = 255 - final
# Save
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, result)
return str(output_path)
def remove_dangling_lines(self, image_path, file_name="dangling_lines_removed.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
gray = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
# Threshold to binary mask (black lines = True, white = False)
binary = gray < 128 # black parts (lines/dangling strokes)
binary = binary.astype(bool)
# Label connected components
labeled = label(binary, connectivity=2)
# Remove small connected components (dangling lines, fragments)
cleaned = remove_small_objects(labeled, min_size=500) # Adjust min_size as needed
# Convert back to mask (255 = black lines kept, 255 background = white)
final_mask = (cleaned > 0).astype(np.uint8) * 255
# Invert mask to match original layout: black lines on white background
final_image = 255 - final_mask
# Save result
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, final_image)
return output_path
def remove_diagonal_lines(self, image_path, file_name="remove_diagonal_lines.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
# Read the image
img = cv2.imread(image_path)
# Convert to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# Create binary image (black lines on white background)
_, binary = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY_INV)
# Create kernels for detecting horizontal and vertical lines
# Adjust kernel size based on your image - larger for thicker lines
kernel_length = max(gray.shape[0], gray.shape[1]) // 30
# Horizontal kernel
horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (kernel_length, 1))
# Vertical kernel
vertical_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (1, kernel_length))
# Detect horizontal lines
horizontal_lines = cv2.morphologyEx(binary, cv2.MORPH_OPEN, horizontal_kernel, iterations=2)
# Detect vertical lines
vertical_lines = cv2.morphologyEx(binary, cv2.MORPH_OPEN, vertical_kernel, iterations=2)
# Combine horizontal and vertical lines
rect_lines = cv2.addWeighted(horizontal_lines, 1, vertical_lines, 1, 0)
# Create final result - white background with black rectangular lines only
result = np.ones_like(gray) * 255 # White background
result[rect_lines > 0] = 0 # Black lines where rectangular lines were detected
# Save result
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, result)
return output_path
def thick_black(self, image_path, thickness=20, file_name="thick_black.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
# Load image
img = cv2.imread(image_path)
# Convert to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# Create a binary mask where black pixels are 1 (foreground)
_, binary = cv2.threshold(gray, 10, 255, cv2.THRESH_BINARY_INV)
# Define kernel size based on desired thickness
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (thickness, thickness))
# Dilate the black areas
dilated = cv2.dilate(binary, kernel, iterations=1)
# Invert back so black is 0 again
# result_mask = cv2.bitwise_not(dilated)
# Apply mask on original image
result = img.copy()
result[np.where(dilated == 255)] = (0, 0, 0)
# Save result
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, result)
return output_path
def to_int_box(self, line):
return map(int, line[0]) # Works for both Hough and LSD formats
def remove_diagonal_lines_and_set_white(self, image_path, file_name="remove_diagonal_lines_and_set_white.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
# Load image
image = cv2.imread(image_path)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# Edge detection
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
blurred = cv2.GaussianBlur(gray, (3, 3), 0)
edges = cv2.Canny(blurred, 50, 150, apertureSize=3)
# Dilate to connect broken segments
kernel = np.ones((2, 2), np.uint8)
edges = cv2.dilate(edges, kernel, iterations=1)
# More sensitive Hough transform
# HoughLinesP_lines = cv2.HoughLinesP(edges, 1, np.pi / 180, threshold=30, minLineLength=5, maxLineGap=10)
# Detect lines using Hough Transform
lsd = cv2.createLineSegmentDetector(0)
lines, _, _, _ = lsd.detect(gray)
# Copy image to edit
output = image.copy()
combined_lines = []
if lines is not None:
combined_lines.extend(lines)
# if HoughLinesP_lines is not None:
# combined_lines.extend(HoughLinesP_lines)
if combined_lines is not None:
for line in combined_lines:
x1, y1, x2, y2 = self.to_int_box(line) # Convert float to int
# Calculate angle
angle = np.abs(np.arctan2(y2 - y1, x2 - x1) * 180.0 / np.pi)
# Filter out horizontal and vertical lines
if (80 < angle < 100) or (170 < angle < 190) or angle < 10 or angle > 350:
continue
else:
# Get bounding box with padding
padding = 2
xmin = min(x1, x2) - padding
xmax = max(x1, x2) + padding
ymin = min(y1, y2) - padding
ymax = max(y1, y2) + padding
# Draw white rectangle (erase diagonal line)
cv2.rectangle(output, (xmin, ymin), (xmax, ymax), (255, 255, 255), thickness=-1)
# Save cleaned image
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, output)
return output_path
def remove_small_regions(self, image_path, file_name="remove_small_regions.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
# Load image in grayscale
img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
visual = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR) # For debugging with colored rectangles
if img is None:
raise FileNotFoundError(f"Could not load image: {image_path}")
height_, width_ = img.shape
min_area = height_ * width_ * self.config.min_area_ratio
# Threshold: make black = foreground
_, binary = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY_INV)
# Label connected regions
labeled = measure.label(binary)
regions = measure.regionprops(labeled)
# Create clean mask (copy of original binary)
clean_mask = np.copy(binary)
for region in regions:
area = region.area
minr, minc, maxr, maxc = region.bbox
width = maxc - minc
height = maxr - minr
# Bounding box filter
if width < width_ * self.config.min_width_ratio and height < height_ * self.config.min_height_ratio:
if (width/width_) < 0.9 and (height/height_) < 0.9:
clean_mask[labeled == region.label] = 0 # Remove small region
cv2.rectangle(visual, (minc, minr), (maxc, maxr), (0, 0, 255), 2)
continue
# Crop and analyze region for line orientation
region_crop = binary[minr:maxr, minc:maxc]
edges = cv2.Canny(region_crop, 50, 150, apertureSize=3)
lines = cv2.HoughLinesP(edges, 1, np.pi / 180, threshold=30, minLineLength=10, maxLineGap=5)
if lines is not None:
for line in lines:
x1, y1, x2, y2 = line[0]
angle = np.abs(np.arctan2(y2 - y1, x2 - x1) * 180.0 / np.pi)
# length = np.sqrt((x2 - x1)**2 + (y2 - y1)**2)
line_width = abs(x2 - x1)
line_height = abs(y2 - y1)
if line_height < height_ * self.config.min_height_ratio and line_width < width_ * self.config.min_width_ratio:
break
else:
# Only runs if no 'break' occurred
# If no qualifying line found, remove region
clean_mask[labeled == region.label] = 0
cv2.rectangle(visual, (minc, minr), (maxc, maxr), (0, 255, 255), 2)
elif width < width_ * self.config.min_width_ratio and height < height_ * self.config.min_height_ratio:
# No lines, remove region
clean_mask[labeled == region.label] = 0
cv2.rectangle(visual, (minc, minr), (maxc, maxr), (255, 0, 0), 2)
# Save debug visualization
output_path = self.get_output_path(output_folder, f"debug_{file_name}")
cv2.imwrite(output_path, visual)
# Invert back to original format: black lines on white
cleaned = cv2.bitwise_not(clean_mask)
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, cleaned)
return output_path
def thin_black(self, image_path, file_name="thin_black.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
# Load the image (replace 'debug_dilated.jpg' with your actual file path if needed)
img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
# Check if the image loaded correctly
if img is None:
raise ValueError("Image not loaded. Check the file path.")
# Threshold to binary (invert if lines are black on white)
_, binary = cv2.threshold(img, 128, 255, cv2.THRESH_BINARY_INV)
# Perform thinning to reduce to 1-pixel lines
try:
# Use Zhang-Suen thinning if opencv-contrib is installed
thinned = cv2.ximgproc.thinning(binary)
except AttributeError:
# Fallback: Morphological skeletonization
skel = np.zeros(binary.shape, np.uint8)
element = cv2.getStructuringElement(cv2.MORPH_CROSS, (3, 3))
while True:
eroded = cv2.erode(binary, element)
temp = cv2.dilate(eroded, element)
temp = cv2.subtract(binary, temp)
skel = cv2.bitwise_or(skel, temp)
binary = eroded.copy()
if cv2.countNonZero(binary) == 0:
break
thinned = skel
# Invert back if needed (for white lines on black background)
thinned = 255 - thinned
# Save result
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, thinned)
return output_path
def thin_lines_direct(self, image_path, file_name="thin_lines_direct.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
# Read image
img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
if img is None:
raise ValueError("Could not load image")
# Convert to binary (0 = black lines, 255 = white background)
_, binary = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)
# Create result image (start with white background)
result = np.full_like(binary, 255) # All white
height, width = binary.shape
print("Processing thick lines...")
# Method 1: Scan rows - for each thick horizontal segment, keep only bottom pixel
print("Step 1: Thinning horizontal segments...")
for row in range(height):
col = 0
while col < width:
# If we hit a black pixel
if binary[row, col] == 0: # Black pixel
# Find the end of this horizontal segment
start_col = col
while col < width and binary[row, col] == 0:
col += 1
end_col = col - 1
# For this horizontal segment, check if it's part of a thick vertical region
segment_width = end_col - start_col + 1
if segment_width >= 1: # Any horizontal segment
# Check how thick this region is vertically at the middle
mid_col = (start_col + end_col) // 2
# Find vertical thickness at this point
thickness = self.get_vertical_thickness(binary, row, mid_col)
if thickness > 1:
# This is part of a thick region - keep only the bottom pixel
bottom_row = row + thickness - 1
if bottom_row < height:
result[bottom_row, start_col:end_col+1] = 0 # Draw black line
else:
# Already thin - keep as is
result[row, start_col:end_col+1] = 0
else:
col += 1
# Save step 1
# cv2.imwrite(f'{self.config.output_folder}/step1_horizontal_thinned.png', result)
# Method 2: Scan columns - for each thick vertical segment, keep only right pixel
print("Step 2: Thinning vertical segments...")
# Start fresh for vertical processing
result_v = np.full_like(binary, 255) # All white
for col in range(width):
row = 0
while row < height:
# If we hit a black pixel
if binary[row, col] == 0: # Black pixel
# Find the end of this vertical segment
start_row = row
while row < height and binary[row, col] == 0:
row += 1
end_row = row - 1
segment_height = end_row - start_row + 1
if segment_height >= 1: # Any vertical segment
# Check how thick this region is horizontally at the middle
mid_row = (start_row + end_row) // 2
# Find horizontal thickness at this point
thickness = self.get_horizontal_thickness(binary, mid_row, col)
if thickness > 1:
# This is part of a thick region - keep only the right pixel
right_col = col + thickness - 1
if right_col < width:
result_v[start_row:end_row+1, right_col] = 0 # Draw black line
else:
# Already thin - keep as is
result_v[start_row:end_row+1, col] = 0
else:
row += 1
# Save step 2
# cv2.imwrite(f'{self.config.output_folder}/step2_vertical_thinned.png', result_v)
# Method 3: Combine both results
print("Step 3: Combining results...")
final_result = cv2.bitwise_and(result, result_v) # Keep both thin lines
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, final_result)
return output_path
def get_vertical_thickness(self, binary, start_row, col):
"""Get the vertical thickness of a black region starting from start_row, col"""
height = binary.shape[0]
thickness = 0
row = start_row
while row < height and binary[row, col] == 0: # Black pixel
thickness += 1
row += 1
return thickness
def get_horizontal_thickness(self, binary, row, start_col):
"""Get the horizontal thickness of a black region starting from row, start_col"""
width = binary.shape[1]
thickness = 0
col = start_col
while col < width and binary[row, col] == 0: # Black pixel
thickness += 1
col += 1
return thickness
def remove_diagonal_only_cells(self, image_path, file_name="remove_diagonal_only_cells.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
# Load the image in grayscale
img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
if img is None:
raise ValueError("Unable to load the image. Check the file path.")
# Threshold to binary (invert if lines are black on white background)
_, binary = cv2.threshold(img, 128, 255, cv2.THRESH_BINARY_INV)
# Pad image to handle border cells easily
padded = np.pad(binary, pad_width=1, mode='constant', constant_values=0)
rows, cols = binary.shape
output = padded.copy()
# Scan each cell (excluding padding)
for r in range(1, rows + 1):
for c in range(1, cols + 1):
if padded[r, c] == 255: # Assuming white (255) represents active cells/lines
# Get 8 neighbors
neighbors = {
'top_left': padded[r-1, c-1],
'top': padded[r-1, c],
'top_right': padded[r-1, c+1],
'left': padded[r, c-1],
'right': padded[r, c+1],
'bottom_left': padded[r+1, c-1],
'bottom': padded[r+1, c],
'bottom_right': padded[r+1, c+1]
}
# Helper: Count active neighbors (255)
active_count = sum(1 for v in neighbors.values() if v == 255)
# Conditions as specified:
# 1) Only top-left and bottom-right
cond1 = (neighbors['top_left'] == 255 and neighbors['bottom_right'] == 255 and
active_count == 2)
# 2) Only top-left
cond2 = (neighbors['top_left'] == 255 and active_count == 1)
# 3) Only bottom-right
cond3 = (neighbors['bottom_right'] == 255 and active_count == 1)
# 4) Only top-right and bottom-left
cond4 = (neighbors['top_right'] == 255 and neighbors['bottom_left'] == 255 and
active_count == 2)
# 5) Only top-right
cond5 = (neighbors['top_right'] == 255 and active_count == 1)
# 6) Only bottom-left
cond6 = (neighbors['bottom_left'] == 255 and active_count == 1)
# Remove cell if any condition matches (set to 0)
if cond1 or cond2 or cond3 or cond4 or cond5 or cond6:
output[r, c] = 0
# Remove padding and invert back to original style (black lines on white)
cleaned = output[1:-1, 1:-1]
result = cv2.bitwise_not(cleaned)
# Save the result
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, result)
return output_path
def remove_small_continuity_components(
self,
image_path,
file_name="remove_small_continuity_components.jpg",
output_folder=None,
):
output_folder = output_folder or self.config.output_folder
# Load the image in grayscale
img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
if img is None:
raise ValueError("Unable to load the image. Check the file path.")
height, width = img.shape
min_height = height * self.config.min_height_ratio
min_width = width * self.config.min_width_ratio
# Threshold to binary (invert if lines are black on white background)
_, binary = cv2.threshold(img, 128, 255, cv2.THRESH_BINARY_INV)
# Perform connected component labeling (8-connectivity)
num_labels, labels, stats, _ = cv2.connectedComponentsWithStats(binary, connectivity=8)
# Create output copies
cleaned_output = binary.copy()
debug_output = cv2.cvtColor(binary.copy(), cv2.COLOR_GRAY2BGR) # For visualizing removed components
for label in tqdm(range(1, num_labels), desc="Processing labels"):
x, y, w, h, area = stats[label]
# Filter out small components based on width and height
if h < min_height and w < min_width:
cleaned_output[labels == label] = 0
debug_output[labels == label] = [0, 0, 255] # Mark removed components in red
# Invert back to original style
final_result = cv2.bitwise_not(cleaned_output)
# Save the final and debug outputs
output_path = self.get_output_path(output_folder, file_name)
debug_path = self.get_output_path(output_folder, file_name.replace(".jpg", "_debug.jpg"))
cv2.imwrite(output_path, final_result)
cv2.imwrite(debug_path, debug_output)
return output_path
def connect_horizontal_vertical_gaps(self, image_path, file_name='connected_output.jpg', output_folder=None):
output_folder = output_folder or self.config.output_folder
image = cv2.imread(image_path)
height, width = image.shape[:2]
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
edges = cv2.Canny(gray, 50, 150, apertureSize=3)
# Detect all lines
lines = cv2.HoughLinesP(edges, 1, np.pi / 180, threshold=50, minLineLength=30, maxLineGap=10)
output = image.copy()
def angle_of_line(x1, y1, x2, y2):
return abs(math.degrees(math.atan2(y2 - y1, x2 - x1)))
# Filter for only horizontal (≈0°) and vertical (≈90°) lines
filtered_lines = []
if lines is not None:
for line in lines:
x1, y1, x2, y2 = line[0]
angle = angle_of_line(x1, y1, x2, y2)
min_width = 0
min_height = 0
if angle < 5:
line_width = abs(x2 - x1)
if line_width >= min_width:
filtered_lines.append([x1, y1, x2, y2])
elif 85 < angle < 95:
line_height = abs(y2 - y1)
if line_height >= min_height:
filtered_lines.append([x1, y1, x2, y2])
# Merge similar lines (if needed)
merged_lines = []
used = [False] * len(filtered_lines)
horizontal_alignment_threshold = 5
horizontal_distance_threshold = width * self.config.min_width_ratio
vertical_alignment_threshold = 5
vertical_distance_threshold = height * self.config.min_height_ratio
overlap_allowance = 10
for i in range(len(filtered_lines)):
if used[i]:
continue
x1a, y1a, x2a, y2a = filtered_lines[i]
merged = [x1a, y1a, x2a, y2a]
used[i] = True
for j in range(i + 1, len(filtered_lines)):
if used[j]:
continue
x1b, y1b, x2b, y2b = filtered_lines[j]
# Check if both are horizontal
if abs(y1a - y2a) < horizontal_alignment_threshold and abs(y1b - y2b) < horizontal_alignment_threshold and abs(y1a - y1b) < horizontal_distance_threshold:
if max(x1a, x2a) >= min(x1b, x2b) - overlap_allowance or max(x1b, x2b) >= min(x1a, x2a) - overlap_allowance:
merged = [
min(merged[0], merged[2], x1b, x2b),
y1a,
max(merged[0], merged[2], x1b, x2b),
y1a
]
used[j] = True
# Check if both are vertical
elif abs(x1a - x2a) < vertical_alignment_threshold and abs(x1b - x2b) < vertical_alignment_threshold and abs(x1a - x1b) < vertical_distance_threshold:
if max(y1a, y2a) >= min(y1b, y2b) - overlap_allowance or max(y1b, y2b) >= min(y1a, y2a) - overlap_allowance:
merged = [
x1a,
min(merged[1], merged[3], y1b, y2b),
x1a,
max(merged[1], merged[3], y1b, y2b)
]
used[j] = True
merged_lines.append(merged)
# Draw merged lines
for x1, y1, x2, y2 in merged_lines:
cv2.line(output, (x1, y1), (x2, y2), (0, 0, 0), 20)
# Save the result
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, output)
return output_path
def detect_small_objects_and_set_white(self, image_path, file_name="detect_small_objects_and_set_white.jpg", output_folder=None):
output_folder = output_folder or self.config.output_folder
# Load image
image = cv2.imread(image_path)
height, width = image.shape[:2]
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# Threshold to binary
_, binary = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY_INV)
# Find contours (external only or all)
contours, _ = cv2.findContours(binary, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
# Draw bounding boxes
output = image.copy()
for cnt in contours:
x, y, w, h = cv2.boundingRect(cnt)
if h < height * self.config.min_height_ratio and w < width * self.config.min_width_ratio:
cv2.rectangle(output, (x, y), (x + w, y + h), (255, 255, 255), -1)
# Save output
output_path = self.get_output_path(output_folder, file_name)
cv2.imwrite(output_path, output)
return output_path
|