fixes to split photos

scripts/split_photos.py

@@ -1,3 +1,4 @@
+import argparse
 import sys
 import cv2
 import numpy as np
@@ -36,33 +37,21 @@ def estimate_background_color(image, sample_points=5):
 
     return np.median(colors, axis=0)
 
-def auto_rotate(image, angle_threshold=10):
+def auto_rotate(image, angle_threshold=1):
     gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
-    ret, binary = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)
-    contours, _ = cv2.findContours(binary, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+    edges = cv2.Canny(gray, 50, 150, apertureSize=3)
+    lines = cv2.HoughLines(edges, 1, np.pi / 180, 200)
 
-    if len(contours) == 0:
+    if lines is None:
         return image
 
-    largest_contour = max(contours, key=cv2.contourArea)
-    mu = cv2.moments(largest_contour)
+    # compute the median angle of the lines
+    angles = []
+    for rho, theta in lines[:, 0]:
+        angles.append((theta * 180) / np.pi - 90)
     
-    if mu["m00"] == 0:
-        return image
+    angle = np.median(angles)
     
-    x_centroid = int(mu["m10"] / mu["m00"])
-    y_centroid = int(mu["m01"] / mu["m00"])
-
-    coords = np.column_stack(np.where(binary > 0))
-    u, _, vt = np.linalg.svd(coords - np.array([[y_centroid, x_centroid]]), full_matrices=False)
-
-    angle = np.arctan2(u[1, 0], u[0, 0]) * 180 / np.pi
-
-    if angle < -45:
-        angle = -(90 + angle)
-    else:
-        angle = -angle
-
     if abs(angle) < angle_threshold:
         return image
 
@@ -73,6 +62,7 @@ def auto_rotate(image, angle_threshold=10):
 
 
 
+
 def crop_borders(image, border_color, tolerance=30):
     mask = cv2.inRange(image, border_color - tolerance, border_color + tolerance)
     
@@ -112,23 +102,15 @@ def split_photos(input_file, output_directory, tolerance=30, min_area=10000, min
         print(f"Saved {output_path}")
 
 if __name__ == "__main__":
-    if len(sys.argv) < 2:
-        print("Usage: python3 split_photos.py <input_file> <output_directory> [tolerance] [min_area] [min_contour_area] [angle_threshold] [border_size]")
-        print("\nParameters:")
-        print("  <input_file>       - The input scanned image containing multiple photos.")
-        print("  <output_directory> - The directory where the result images should be placed.")
-        print("  [tolerance]        - Optional. Determines the range of color variation around the estimated background color (default: 30).")
-        print("  [min_area]         - Optional. Sets the minimum area threshold for a photo (default: 10000).")
-        print("  [min_contour_area] - Optional. Sets the minimum contour area threshold for a photo (default: 500).")
-        print("  [angle_threshold]  - Optional. Sets the minimum absolute angle required for the image to be rotated (default: 10).")
-        print("  [border_size]      - Optional. Sets the size of the border added and removed to prevent white borders in the output (default: 0).")
-        sys.exit(1)
+    parser = argparse.ArgumentParser(description="Split photos in an image")
+    parser.add_argument("input_file", help="The input scanned image containing multiple photos.")
+    parser.add_argument("output_directory", help="The directory where the result images should be placed.")
+    parser.add_argument("--tolerance", type=int, default=30, help="Determines the range of color variation around the estimated background color (default: 30).")
+    parser.add_argument("--min_area", type=int, default=10000, help="Sets the minimum area threshold for a photo (default: 10000).")
+    parser.add_argument("--min_contour_area", type=int, default=500, help="Sets the minimum contour area threshold for a photo (default: 500).")
+    parser.add_argument("--angle_threshold", type=int, default=10, help="Sets the minimum absolute angle required for the image to be rotated (default: 10).")
+    parser.add_argument("--border_size", type=int, default=0, help="Sets the size of the border added and removed to prevent white borders in the output (default: 0).")
+    
+    args = parser.parse_args()
 
-    input_file = sys.argv[1]
-    output_directory = sys.argv[2]
-    tolerance = int(sys.argv[6]) if len(sys.argv) > 3 else 20
-    min_area = int(sys.argv[8]) if len(sys.argv) > 4 else 8000
-    min_contour_area = int(sys.argv[10]) if len(sys.argv) > 5 else 500
-    angle_threshold = int(sys.argv[4]) if len(sys.argv) > 6 else 60
-    border_size = int(sys.argv[12]) if len(sys.argv) > 7 else 0
-    split_photos(input_file, output_directory, tolerance=tolerance, min_area=min_area, min_contour_area=min_contour_area, angle_threshold=angle_threshold, border_size=border_size)
+    split_photos(args.input_file, args.output_directory, tolerance=args.tolerance, min_area=args.min_area, min_contour_area=args.min_contour_area, angle_threshold=args.angle_threshold, border_size=args.border_size)

src/main/resources/templates/other/extract-image-scans.html

@@ -18,7 +18,7 @@
               <div th:replace="~{fragments/common :: fileSelector(name='fileInput', multiple=false, accept='image/*, application/pdf')}"></div>
               <div class="form-group">
                 <label for="angleThreshold" th:text="#{ScannerImageSplit.selectText.1}"></label>
-                <input type="number" class="form-control" id="angleThreshold" name="angle_threshold" value="5">
+                <input type="number" class="form-control" id="angleThreshold" name="angle_threshold" value="10">
                 <small id="angleThresholdHelp" class="form-text text-muted" th:text="#{ScannerImageSplit.selectText.2}"></small>
               </div>
               <div class="form-group">