Merge pull request #10 from sivakumar-mahalingam/dev

Dev _get_roi

Author: sivakumar-mahalingam

README.md

@@ -137,7 +137,8 @@ Now, based on the example of a national passport, let us take a closer look at t
 - [ ] Support pdf as input
 - [x] Function to return mrz text as output
 - [ ] Bulk process
-- [ ] Image Enhancement
+- [ ] Add function parameter - Image Enhancement
+- [ ] Add function parameter - Image Enhancement Model
 - [ ] Train Tesseract model with additional data
 - [x] Add function parameter - include_checkdigit
 - [ ] Add function - get_mrz_image

fastmrz/fastmrz.py

@@ -56,15 +56,16 @@ def _get_roi(self, output_data, image_path):
         y_end = min(image.shape[0], y + h + padding)
 
         roi_arr = image[y_start:y_end, x_start:x_end].copy()
-        roi_arr = cv2.convertScaleAbs(roi_arr, alpha=1.25, beta=-50)
+        # roi_arr = cv2.convertScaleAbs(roi_arr, alpha=1.25, beta=-50)
 
         # Apply additional preprocessing to ROI before OCR
         roi_gray = cv2.cvtColor(roi_arr, cv2.COLOR_BGR2GRAY)
 
-        kernel = np.ones((1,1), np.uint8)
-        roi_dilate = cv2.dilate(roi_gray, kernel, iterations=1)
-        roi_gaussianblur = cv2.GaussianBlur(roi_dilate, (5,5), 0)
-        roi_threshold = cv2.threshold(roi_gaussianblur, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]
+        # kernel = np.ones((1, 1), np.uint8)
+        # roi_dilate = cv2.dilate(roi_gray, kernel, iterations=1)
+        # roi_dilate = cv2.erode(roi_dilate, kernel, iterations=1)
+
+        roi_threshold = cv2.threshold(roi_gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]
 
         # Configure pytesseract parameters for better MRZ recognition
         custom_config = r'--oem 3 --psm 6'