Added the updated logic for Water Bill OCR

openbharatocr/ocr/water_bill.py

@@ -3,11 +3,24 @@
 from PIL import Image
 from datetime import datetime
 
+def preprocess_for_bold_text(image):
 
-def extract_name(input):
+    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
 
-    regex = r"(?:Name\s*[:\s]?\s*)(.*?)(?:\bConsumer\b|['/]|$)"
-    match = re.search(regex, input, re.IGNORECASE)
+    kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (1, 1))
+    opening = cv2.morphologyEx(gray, cv2.MORPH_OPEN, kernel)
+    contrast = cv2.addWeighted(opening, 2, opening, -0.5, 0)
+
+    _, binary = cv2.threshold(contrast, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)
+    sharpened = cv2.filter2D(
+        binary, -1, np.array([[0, -1, 0], [-1, 5, -1], [0, -1, 0]])
+    )
+
+    return sharpened
+
+def extract_name(input):
+    regex = r"Name:\s*(.*?)(?:\.\s|(?=\n))"
+    match = re.search(regex, input)
     name = match.group(1).strip() if match else ""
 
     if name == "":
@@ -16,85 +29,97 @@ def extract_name(input):
         name = match.group(1).strip() if match else ""
 
     return name
-
-
+
+def extract_bill_amount(input):
+    regex = r"Bill Amount \(Rs\.\)\s*:? (\d+)"
+    match = re.search(regex, input, re.IGNORECASE)
+    bill_amount = match.group(1).strip() if match else ""
+    return bill_amount
+
+def extract_meter_number(input):
+    regex = r"Meter No\.\s*:\s*(\d+|NA)"
+    match = re.search(regex, input, re.IGNORECASE)
+    meter_number = match.group(1).strip() if match else ""
+    return meter_number
+
 def extract_all_dates(input):
-
-    regex = r"\b(\d{1,2}[/\-.](?:\d{2}|\d{4}|\w{3})[/\-.]\d{2,4})\b"
+    regex = r"\b(\d{1,2}-[A-Z]{3}-\d{4})\b"
     dates = re.findall(regex, input)
     formatted_dates = []
     for date in dates:
         try:
-            formatted_date = datetime.strptime(date, "%d/%m/%Y")
+            formatted_date = datetime.strptime(date, "%d-%b-%Y")
+            formatted_dates.append(formatted_date)
         except ValueError:
-            try:
-                formatted_date = datetime.strptime(date, "%d-%m-%Y")
-            except ValueError:
-                formatted_date = datetime.strptime(date, "%d-%b-%Y")
-        formatted_dates.append(formatted_date)
-
+            continue
     sorted_dates = sorted(formatted_dates)
-    sorted_dates_str = [date.strftime("%d/%m/%Y") for date in sorted_dates]
-
+    sorted_dates_str = [date.strftime("%d-%m-%Y") for date in sorted_dates]
     return sorted_dates_str
 
-
 def extract_phone(input):
-
-    regex = r"[6789]\d{9}"
+    regex = r"Mobile No\.\s*:\s*(\d+)"
     match = re.search(regex, input)
-    phone = match.group(0) if match else ""
-
+    phone = match.group(1).strip() if match else ""
     return phone
 
-
 def extract_address(input):
-
-    regex = r"Address:\s*(.*?)(?=\b[A-Z][a-zA-Z\s]*:|\b[A-Z][a-zA-Z\s]*$)"
+    regex = r"Address\s*:\s*(.*?)(?=\s*[A-Z][a-zA-Z\s]*:|$)"
     match = re.search(regex, input, re.DOTALL | re.IGNORECASE)
     address = match.group(1).strip() if match else ""
-
     return address
-
-
-def extract_meter_type(input):
-    regex = r"Meter\s*Type\s*.DJB.Pvt.:\s*([A-Z]+)\s"
+
+def extract_mr_code(input):
+    regex = r"Zone/MR\s*Code:\s*([A-Z0-9/]+\s*[A-Z0-9/]*)"
     match = re.search(regex, input, re.IGNORECASE)
-    meter_type = match.group(1).strip() if match else ""
-
-    return meter_type
+    mr_code = match.group(1).strip() if match else ""
+    return mr_code
 
-
-def extract_bill_amount(input):
-
-    regex = r"total\sdue\sdate\s*(\d+)\n"
+def extract_area_code(input):
+    regex = r"Area Code\s*:\s*([\w/-]+)"
     match = re.search(regex, input, re.IGNORECASE)
-    bill_amount = match.group(1).strip() if match else ""
+    area_code = match.group(1).strip() if match else ""
+    return area_code
 
-    if bill_amount == "":
-        regex = r".Rs.\s*(\d+)\n"
-        matches = re.findall(regex, input, re.IGNORECASE)
-        bill_amount = matches[-1] if len(matches) > 0 else ""
+def extract_bill_number(input):
+    regex = r"Bill No\.\s*(?::\s*)?(\d+)"
+    match = re.search(regex, input)
+    bill_number = match.group(1).strip() if match else ""
+    return bill_number
 
-    return bill_amount
+def extract_govt_body(input):
+    regex = r"Delhi Jal Board"
+    match = re.search(regex, input, re.IGNORECASE)
+    govt_body = match.group(0).strip() if match else "Unknown"
+    return govt_body
 
+def extract_bill_date(input):
+    regex = r"Bill Date\s*:? (\d{2}-[A-Z]{3}-\d{4})"
+    match = re.search(regex, input, re.IGNORECASE)
+    bill_date = match.group(1).strip() if match else ""
+    return bill_date
 
+def extract_bill_due_date(input):
+    regex = r"Bill Due Date\s*:? (\d{2}-[A-Z]{3}-\d{4})"
+    match = re.search(regex, input, re.IGNORECASE)
+    bill_due_date = match.group(1).strip() if match else ""
+    return bill_due_date
+
 def extract_water_bill_details(image_path):
-
     image = Image.open(image_path)
     extracted_text = pytesseract.image_to_string(image)
 
     name = extract_name(extracted_text)
-
     dates = extract_all_dates(extracted_text)
-    bill_date = dates[0] if len(dates) > 0 else ""
-    due_date = dates[1] if len(dates) > 1 else ""
-
+    bill_date = extract_bill_date(extracted_text)
+    due_date = extract_bill_due_date(extracted_text)
     address = extract_address(extracted_text)
     phone = extract_phone(extracted_text)
-
-    meter_type = extract_meter_type(extracted_text)
     bill_amount = extract_bill_amount(extracted_text)
+    mr_code = extract_mr_code(extracted_text)
+    area_code = extract_area_code(extracted_text)
+    meter_number = extract_meter_number(extracted_text)
+    bill_number = extract_bill_number(extracted_text)
+    govt_body = extract_govt_body(extracted_text)
 
     return {
         "Name": name,
@@ -103,9 +128,13 @@ def extract_water_bill_details(image_path):
         "Bill Date": bill_date,
         "Due Date": due_date,
         "Address": address,
-        "Meter Type": meter_type,
+        "Zone/MR Code": mr_code,
+        "Area Code": area_code,
+        "Meter Number": meter_number,
+        "Bill Number": bill_number,
+        "Source/Govt Body Name": govt_body,
     }
 
-
 def water_bill(image_path):
     return extract_water_bill_details(image_path)
+