Sanjayraju30 commited on
Commit
7dd3534
·
verified ·
1 Parent(s): 8c50e18

Update ocr_engine.py

Browse files
Files changed (1) hide show
  1. ocr_engine.py +13 -5
ocr_engine.py CHANGED
@@ -9,15 +9,23 @@ def extract_weight_from_image(pil_img):
9
  try:
10
  img = np.array(pil_img)
11
 
 
12
  gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
13
- resized = cv2.resize(gray, None, fx=2, fy=2, interpolation=cv2.INTER_CUBIC)
14
- blurred = cv2.GaussianBlur(resized, (5, 5), 0)
15
- _, thresh = cv2.threshold(blurred, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
16
 
17
- result = reader.readtext(thresh, detail=0)
18
- text = " ".join(result)
 
 
 
 
 
 
 
 
 
19
  print("OCR Text:", text)
20
 
 
21
  match = re.search(r"\b\d{2,4}\.?\d{0,2}\b", text)
22
  if match:
23
  return match.group(), 95.0
 
9
  try:
10
  img = np.array(pil_img)
11
 
12
+ # Step 1: Convert to grayscale
13
  gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
 
 
 
14
 
15
+ # Step 2: Apply adaptive threshold to handle lighting
16
+ thresh = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
17
+ cv2.THRESH_BINARY_INV, 11, 2)
18
+
19
+ # Step 3: Dilate to make digits thicker
20
+ kernel = np.ones((2, 2), np.uint8)
21
+ dilated = cv2.dilate(thresh, kernel, iterations=1)
22
+
23
+ # Step 4: OCR on the preprocessed image
24
+ result = reader.readtext(dilated, detail=0)
25
+ text = " ".join(result).strip()
26
  print("OCR Text:", text)
27
 
28
+ # Step 5: Match numeric values like 52.30 or 003.25
29
  match = re.search(r"\b\d{2,4}\.?\d{0,2}\b", text)
30
  if match:
31
  return match.group(), 95.0