pycognaize.common.utils.stick_word_boxes

stick_word_boxes(box_coord, img_bytes, padding=1)[source]
Stick boxes with OpenCV
  1. Stick by y (height), add “2 pixel” padding in bottom and right

    Get bigger sequence of 255 < vectors_y < 0 –> bigger_sequence new_top = min(bigger_sequence) new_bottom = max(bigger_sequence)

  2. stick by x (width), add “2 pixel” padding in bottom

    Get numpy array of 255 < vectors_x < 0 –> min_max_x new_left = min(min_max_x) new_right = max(min_max_x)

Parameters:

  • box_coord (List[dict]) – List of dictionaries representing boxes with the following format: ({‘left’: <int>, ‘right’: <int>, ‘top’: <int>, ‘bottom’: <int>})

  • img_bytes (bytes) – The original image to which the coordinates correspond to

  • padding – Expand the final boxes with this amount of pixels on each four sides

Returns: