diff --git a/OCR.py b/OCR.py index 76c551a..06fa919 100644 --- a/OCR.py +++ b/OCR.py @@ -1,6 +1,7 @@ import cv2 from PIL import Image import pytesseract +from wand.image import Image as WandImage card_name_roi = { 'x':44, 'y':32, 'w':360, 'h':40 } @@ -35,7 +36,14 @@ def thick_font(image): file_path = "data/WTR/WTR003.png" -img = cv2.imread(file_path) +with WandImage(filename=file_path) as img2: + width = img2.width + height = img2.height + img2.resize(width*4, height*4) + img2.save(filename="temp/big.jpg") + +big_file_path = "temp/big.jpg" +img = cv2.imread(big_file_path) inverted_image = cv2.bitwise_not(img) cv2.imwrite("temp/inverted.jpg", inverted_image) @@ -55,7 +63,7 @@ dilated_image = thick_font(no_noise) cv2.imwrite("temp/dilated_image.jpg", dilated_image) im = Image.open(file_path) -ocr_result = pytesseract.image_to_string(im) +ocr_result = pytesseract.image_to_string(im, config='psm 11') print("Unaltered:") print (ocr_result) print("===========================") @@ -106,5 +114,5 @@ def testArea(image, x,y,w,h): ocr_result = pytesseract.image_to_string(gray_image) print(ocr_result) -testArea(img, card_name_roi['x'], card_name_roi['y'], card_name_roi['w'], card_name_roi['h']) +testArea(img, card_name_roi['x']*4, card_name_roi['y']*4, card_name_roi['w']*4, card_name_roi['h']*4) cv2.imwrite("temp/markup.jpg", img) \ No newline at end of file