Update ocr_engine.py

ocr_engine.py

@@ -5,62 +5,62 @@ import re
 import logging
 from PIL import Image
 
-# Set up logging
+# Setup logging
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
 
-def preprocess_for_ocr(img):
-    """Apply grayscale, blur, and threshold to prepare image for OCR."""
+
+def preprocess_strong(img):
+    """Sharpen and enhance contrast for blurry weight images."""
     gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
-    blurred = cv2.GaussianBlur(gray, (5, 5), 0)
 
-    # Adaptive threshold
-    thresh = cv2.adaptiveThreshold(
-        blurred, 255,
-        cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
-        cv2.THRESH_BINARY,
-        11, 2
-    )
+    # Resize up for OCR (scale x2)
+    h, w = gray.shape
+    gray = cv2.resize(gray, (w * 2, h * 2), interpolation=cv2.INTER_CUBIC)
+
+    # CLAHE for contrast enhancement
+    clahe = cv2.createCLAHE(clipLimit=3.0, tileGridSize=(8, 8))
+    enhanced = clahe.apply(gray)
+
+    # Strong sharpening
+    kernel = np.array([[0, -1, 0],
+                       [-1, 5, -1],
+                       [0, -1, 0]])
+    sharpened = cv2.filter2D(enhanced, -1, kernel)
+
+    return sharpened
 
-    # Invert to make text white on black
-    inverted = cv2.bitwise_not(thresh)
-    return inverted
 
 def extract_weight_from_image(pil_img):
-    """Extract weight reading from an image using pytesseract."""
+    """Extract weight from an image using multiple Tesseract strategies."""
     try:
-        # Convert PIL to OpenCV
         img = np.array(pil_img)
         img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
 
-        # Preprocess
-        processed_img = preprocess_for_ocr(img)
-
-        # Tesseract config
-        config = r'--oem 3 --psm 7 -c tessedit_char_whitelist=0123456789.'
-
-        # Run OCR
-        text = pytesseract.image_to_string(processed_img, config=config)
+        processed = preprocess_strong(img)
 
-        # Clean text
-        text = text.strip().replace('\n', '').replace(' ', '')
-        text = re.sub(r"[^\d.]", "", text)
+        # OCR configs
+        configs = [
+            r'--oem 3 --psm 6 -c tessedit_char_whitelist=0123456789.',
+            r'--oem 3 --psm 11 -c tessedit_char_whitelist=0123456789.'
+        ]
 
-        # Handle multiple dots
-        if text.count('.') > 1:
-            text = text.replace('.', '', text.count('.') - 1)
+        for config in configs:
+            raw_text = pytesseract.image_to_string(processed, config=config)
+            logging.info(f"[Tesseract Output {config}] Raw text: {raw_text}")
+            cleaned = raw_text.strip().replace('\n', '').replace(' ', '')
+            cleaned = re.sub(r"[^\d.]", "", cleaned)
+            if cleaned.count('.') > 1:
+                cleaned = cleaned.replace('.', '', cleaned.count('.') - 1)
+            if cleaned.startswith('.'):
+                cleaned = '0' + cleaned
 
-        if text.startswith('.'):
-            text = '0' + text
+            if cleaned and re.fullmatch(r"\d*\.?\d*", cleaned):
+                value = float(cleaned)
+                if 0.001 <= value <= 5000:
+                    return str(round(value, 2)), 90.0
 
-        # Validate
-        if text and re.fullmatch(r"\d*\.?\d*", text):
-            value = float(text)
-            if 0.001 <= value <= 5000:
-                return text, 90.0  # Return with fixed confidence
-            else:
-                logging.warning(f"Detected weight out of range: {value}")
         return "Not detected", 0.0
 
     except Exception as e:
-        logging.error(f"OCR error: {str(e)}")
+        logging.error(f"OCR failed: {e}")
         return "Not detected", 0.0