|
| 1 | +"""dojo module. |
| 2 | +
|
| 3 | +1. **Inverted Images** |
| 4 | +2. Rescaling |
| 5 | +3. Binarization |
| 6 | +4. Noise Removal |
| 7 | +5. Dilation and Erosion |
| 8 | +6. Rotation / Deskewing |
| 9 | +7. Removing Borders |
| 10 | +8. Missing Borders |
| 11 | +9. Transparency / Alpha Channel |
| 12 | +""" |
| 13 | + |
| 14 | +from __future__ import annotations |
| 15 | + |
| 16 | +import logging |
| 17 | +import sys |
| 18 | +from copy import copy |
| 19 | +from pathlib import Path |
| 20 | +from typing import TYPE_CHECKING, NoReturn |
| 21 | + |
| 22 | +import cv2 |
| 23 | +from deprecated import deprecated |
| 24 | +from icecream import ic |
| 25 | +from matplotlib import pyplot as plt |
| 26 | + |
| 27 | +if sys.version_info >= (3, 11): |
| 28 | + from typing import Self |
| 29 | +else: |
| 30 | + from typing_extensions import Self |
| 31 | + |
| 32 | +if TYPE_CHECKING: |
| 33 | + import numpy as np |
| 34 | + |
| 35 | +IMG_DIR: Path = Path(__file__).parents[1] / 'generic_data' / 'text_img' |
| 36 | + |
| 37 | + |
| 38 | +def who(cls): |
| 39 | + """Its Class name.""" |
| 40 | + cls.class_name = cls.__name__ |
| 41 | + return cls |
| 42 | + |
| 43 | + |
| 44 | +def add_method(method): |
| 45 | + """Adding method into class.""" |
| 46 | + |
| 47 | + def wrapper(cls): |
| 48 | + """Set method into class.""" |
| 49 | + cls.method = method |
| 50 | + |
| 51 | + |
| 52 | +@deprecated( |
| 53 | + version='1.69.0a5', |
| 54 | + reason='use another implementation updated into dojo20250114.', |
| 55 | +) |
| 56 | +class PreprocessImageOCR: |
| 57 | + """Preprocess Image.""" |
| 58 | + |
| 59 | + def __init__(self, img_path: Path | None = None): |
| 60 | + """Initializer.""" |
| 61 | + self.dpi: float = 80.0 |
| 62 | + self._img_path: Path = None |
| 63 | + self._img_data: np.ndarray = None |
| 64 | + self.img: np.ndarray = None |
| 65 | + self.img_path = img_path |
| 66 | + |
| 67 | + @property |
| 68 | + def img_path(self) -> Path: |
| 69 | + """Imagem file.""" |
| 70 | + return self._img_path |
| 71 | + |
| 72 | + @img_path.setter |
| 73 | + def img_path(self, value: Path) -> NoReturn: |
| 74 | + self._img_path = value |
| 75 | + try: |
| 76 | + self._img_data = plt.imread(self._img_path) |
| 77 | + self.img = copy(self._img_data) |
| 78 | + except AttributeError: |
| 79 | + pass |
| 80 | + logging.info(ic('Image load')) |
| 81 | + logging.debug(ic(self._img_path)) |
| 82 | + |
| 83 | + def save(self, fout: Path | None = None) -> Path: |
| 84 | + """Save current image.""" |
| 85 | + fout = fout or ( |
| 86 | + Path.cwd() / f'{self.img_path.stem}_latest{self.img_path.suffix}' |
| 87 | + ) |
| 88 | + fout = fout.resolve() |
| 89 | + fout.parent.mkdir(exist_ok=True) |
| 90 | + cv2.imwrite(fout, self.img) |
| 91 | + logging.info(ic('Image saved.')) |
| 92 | + logging.debug(ic(fout)) |
| 93 | + return fout |
| 94 | + |
| 95 | + def reset(self) -> Self: |
| 96 | + """Reset to original image.""" |
| 97 | + self.img = copy(self._img_data) |
| 98 | + logging.info(ic('Image reseted.')) |
| 99 | + return self |
| 100 | + |
| 101 | + def display(self, img_path: Path | None = None) -> bool: |
| 102 | + """Display image on screen. |
| 103 | +
|
| 104 | + https://stackoverflow.com/questions/28816046/displaying-different-images-with-actual-size-in-matplotlib-subplot |
| 105 | + """ |
| 106 | + self.img_path = img_path |
| 107 | + |
| 108 | + height, width = self.img.shape[:2] |
| 109 | + |
| 110 | + # What size does the figure need to be in inches to fit the image? |
| 111 | + figsize = width / self.dpi, height / self.dpi |
| 112 | + |
| 113 | + # Create a figure of the right size with |
| 114 | + # one axes that takes up the full figure |
| 115 | + fig = plt.figure(figsize=figsize) |
| 116 | + ax = fig.add_axes([0, 0, 1, 1]) |
| 117 | + |
| 118 | + # Hide spines, ticks, etc. |
| 119 | + ax.axis('off') |
| 120 | + |
| 121 | + # Display the image. |
| 122 | + ax.imshow(self.img, cmap='gray') |
| 123 | + |
| 124 | + plt.show() |
| 125 | + |
| 126 | + |
| 127 | +@who |
| 128 | +class PPIOCR(PreprocessImageOCR): |
| 129 | + """New class.""" |
| 130 | + |
| 131 | + @deprecated( |
| 132 | + version='1.69.0a6', |
| 133 | + reason='use another implementation updated into dojo20250114.', |
| 134 | + ) |
| 135 | + def inverted(self) -> Self: |
| 136 | + """Inverter bit image.""" |
| 137 | + self.img = cv2.bitwise_not(self.img) |
| 138 | + return self |
| 139 | + |
| 140 | + |
| 141 | +if __name__ == '__main__': |
| 142 | + o = PPIOCR(IMG_DIR / 'letter.jpg') |
| 143 | + o.display() |
| 144 | + o.inverted().display() |
0 commit comments