lukas/shenzhen-solitaire
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Made image detection work

Browse Source
This commit is contained in:
Lukas Wölfer
2020-02-05 01:51:30 +01:00
parent 7e18f1db1c
commit c1026c0640
7 changed files with 208 additions and 144 deletions
Download Patch File Download Diff File Expand all files Collapse all files

63
shenzhen_solitaire/card_detection/board_parser.py
View File

@@ -4,34 +4,49 @@ import numpy as np
from .configuration import Configuration
from ..board import Board
from . import card_finder
import cv2
from typing import Iterable, Any, List
import itertools
def parse_board(image: np.ndarray, conf: Configuration) -> Board:
"""Parse a screenshot of the game, using a given configuration"""
fake_adjustments = conf.field_adjustment
fake_adjustments.x -= 5
fake_adjustments.y -= 5
fake_adjustments.h += 10
fake_adjustments.w += 10
row_count = 13
column_count = 8
def grouper(iterable: Iterable[Any], groupsize: int, fillvalue: Any = None) -> Iterable[Any]:
"Collect data into fixed-length chunks or blocks"
args = [iter(iterable)] * groupsize
return itertools.zip_longest(*args, fillvalue=fillvalue)
squares = card_finder.get_field_squares(
image, conf.field_adjustment, count_x=13, count_y=8)
squares = [card_finder.simplify(square)[0] for square in squares]
square_rows = [squares[13 * i:13 * (i + 1)] for i in range(8)]
empty_square = np.full(
shape=(conf.field_adjustment.w,
conf.field_adjustment.h),
fill_value=card_finder.GREYSCALE_COLOR[card_finder.Cardcolor.Background],
dtype=np.uint8)
assert empty_square.shape == squares[0].shape
result: Board = Board()
for row_id, square_row in enumerate(square_rows):
for square in square_row:
fitting_square, _ = card_finder.find_square(
square, [empty_square] + [x[0] for x in conf.catalogue])
if np.array_equal(fitting_square, empty_square):
print("empty")
break
for cat_square, cardtype in conf.catalogue:
if np.array_equal(fitting_square, cat_square):
print(cardtype)
result.field[row_id].append(cardtype)
break
else:
print("did not find image")
image, conf.field_adjustment, count_x=row_count, count_y=column_count
)
grouped_squares = grouper(squares, row_count)
result = Board()
for group_index, square_group in enumerate(grouped_squares):
group_field = []
for index, square in enumerate(square_group):
best_val = None
best_name = None
for template, name in conf.catalogue:
res = cv2.matchTemplate(square, template, cv2.TM_CCOEFF_NORMED)
min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
if best_val is None or max_val > best_val:
best_val = max_val
best_name = name
assert best_name is not None
group_field.append(best_name)
# print(f"\t{best_val}: {best_name}")
# cv2.imshow("Catalogue", cv2.resize(square, (500, 500)))
# cv2.waitKey()
result.field[group_index] = group_field
return result

114
shenzhen_solitaire/card_detection/card_finder.py
View File

@@ -3,25 +3,23 @@
from typing import List, Tuple, Optional, Dict
import enum
import itertools
import numpy as np # type: ignore
import cv2 # type: ignore
import numpy as np
import cv2
from .adjustment import Adjustment, get_square
from ..board import Card, NumberCard, SpecialCard
def _extract_squares(image: np.ndarray,
squares: List[Tuple[int,
int,
int,
int]]) -> List[np.ndarray]:
return [image[square[1]:square[3], square[0]:square[2]].copy()
for square in squares]
def _extract_squares(
image: np.ndarray, squares: List[Tuple[int, int, int, int]]
) -> List[np.ndarray]:
return [
image[square[1] : square[3], square[0] : square[2]].copy() for square in squares
]
def get_field_squares(image: np.ndarray,
adjustment: Adjustment,
count_x: int,
count_y: int) -> List[np.ndarray]:
def get_field_squares(
image: np.ndarray, adjustment: Adjustment, count_x: int, count_y: int
) -> List[np.ndarray]:
"""Return all squares in the field, according to the adjustment"""
squares = []
for index_x, index_y in itertools.product(range(count_y), range(count_x)):
@@ -31,64 +29,43 @@ def get_field_squares(image: np.ndarray,
class Cardcolor(enum.Enum):
"""Relevant colors for different types of cards"""
Bai = (65, 65, 65)
Black = (0, 0, 0)
Red = (22, 48, 178)
Green = (76, 111, 19)
Background = (178, 194, 193)
GREYSCALE_COLOR = {
Cardcolor.Bai: 50,
Cardcolor.Black: 100,
Cardcolor.Red: 150,
Cardcolor.Green: 200,
Cardcolor.Background: 250}
def simplify(image: np.ndarray) -> Tuple[np.ndarray, Dict[Cardcolor, int]]:
"""Reduce given image to the colors in Cardcolor"""
result_image: np.ndarray = np.zeros(
(image.shape[0], image.shape[1]), np.uint8)
result_dict: Dict[Cardcolor, int] = {c: 0 for c in Cardcolor}
for pixel_x, pixel_y in itertools.product(
range(result_image.shape[0]),
range(result_image.shape[1])):
pixel = image[pixel_x, pixel_y]
best_color: Optional[Tuple[Cardcolor, int]] = None
for color in Cardcolor:
mse = sum((x - y) ** 2 for x, y in zip(color.value, pixel))
if not best_color or best_color[1] > mse: #pylint: disable=E1136
best_color = (color, mse)
assert best_color
result_image[pixel_x, pixel_y] = GREYSCALE_COLOR[best_color[0]]
result_dict[best_color[0]] += 1
return (result_image, result_dict)
def _find_single_square(search_square: np.ndarray,
template_square: np.ndarray) -> Tuple[int, Tuple[int, int]]:
def _find_single_square(
search_square: np.ndarray, template_square: np.ndarray
) -> Tuple[int, Tuple[int, int]]:
assert search_square.shape[0] >= template_square.shape[0]
assert search_square.shape[1] >= template_square.shape[1]
best_result: Optional[Tuple[int, Tuple[int, int]]] = None
for margin_x, margin_y in itertools.product(
range(search_square.shape[0], template_square.shape[0] - 1, -1),
range(search_square.shape[1], template_square.shape[1] - 1, -1)):
search_region = search_square[margin_x -
template_square.shape[0]:margin_x, margin_y -
template_square.shape[1]:margin_y]
range(search_square.shape[0], template_square.shape[0] - 1, -1),
range(search_square.shape[1], template_square.shape[1] - 1, -1),
):
search_region = search_square[
margin_x - template_square.shape[0] : margin_x,
margin_y - template_square.shape[1] : margin_y,
]
count = cv2.countNonZero(search_region - template_square)
if not best_result or count < best_result[0]: #pylint: disable=E1136
if not best_result or count < best_result[0]: # pylint: disable=E1136
best_result = (
count,
(margin_x - template_square.shape[0],
margin_y - template_square.shape[1]))
(
margin_x - template_square.shape[0],
margin_y - template_square.shape[1],
),
)
assert best_result
return best_result
def find_square(search_square: np.ndarray,
squares: List[np.ndarray]) -> Tuple[np.ndarray, int]:
def find_square(
search_square: np.ndarray, squares: List[np.ndarray]
) -> Tuple[np.ndarray, int]:
"""Compare all squares in squares with search_square, return best matching one.
Requires all squares to be simplified."""
best_set = False
@@ -104,24 +81,24 @@ def find_square(search_square: np.ndarray,
return (best_square, best_count)
def catalogue_cards(squares: List[np.ndarray]
) -> List[Tuple[np.ndarray, Card]]:
def catalogue_cards(squares: List[np.ndarray]) -> List[Tuple[np.ndarray, Card]]:
"""Run manual cataloging for given squares"""
cv2.namedWindow("Catalogue", cv2.WINDOW_NORMAL)
cv2.waitKey(1)
result: List[Tuple[np.ndarray, Card]] = []
print(
"Card ID is [B]ai, [Z]hong, [F]a, [H]ua, [R]ed, [G]reen, [B]lack")
print("Card ID is [B]ai, [Z]hong, [F]a, [H]ua, [R]ed, [G]reen, [B]lack")
print("Numbercard e.g. R3")
special_card_map = {
'b': SpecialCard.Bai,
'z': SpecialCard.Zhong,
'f': SpecialCard.Fa,
'h': SpecialCard.Hua}
"b": SpecialCard.Bai,
"z": SpecialCard.Zhong,
"f": SpecialCard.Fa,
"h": SpecialCard.Hua,
}
suit_map = {
'r': NumberCard.Suit.Red,
'g': NumberCard.Suit.Green,
'b': NumberCard.Suit.Black}
"r": NumberCard.Suit.Red,
"g": NumberCard.Suit.Green,
"b": NumberCard.Suit.Black,
}
for square in squares:
while True:
cv2.imshow("Catalogue", cv2.resize(square, (500, 500)))
@@ -137,10 +114,11 @@ def catalogue_cards(squares: List[np.ndarray]
continue
if not card_id[1].isdigit():
continue
if card_id[1] == '0':
if card_id[1] == "0":
continue
card_type = NumberCard(number=int(
card_id[1]), suit=suit_map[card_id[0]])
card_type = NumberCard(
number=int(card_id[1]), suit=suit_map[card_id[0]]
)
else:
continue
assert card_type is not None

89
shenzhen_solitaire/card_detection/configuration.py
View File

@@ -4,6 +4,8 @@ import json
from typing import List, Tuple, Dict
import io
import dataclasses
import tempfile
import cv2
import numpy as np
from . import adjustment
@@ -13,15 +15,16 @@ from .. import board
class Configuration:
"""Configuration for solitaire cv"""
ADJUSTMENT_FILE_NAME = 'adjustment.json'
TEMPLATES_DIRECTORY = 'templates'
def __init__(self,
adj: adjustment.Adjustment,
catalogue: List[Tuple[np.ndarray,
board.Card]],
meta: Dict[str,
str]) -> None:
ADJUSTMENT_FILE_NAME = "adjustment.json"
TEMPLATES_DIRECTORY = "templates"
def __init__(
self,
adj: adjustment.Adjustment,
catalogue: List[Tuple[np.ndarray, board.Card]],
meta: Dict[str, str],
) -> None:
self.field_adjustment = adj
self.catalogue = catalogue
self.meta = meta
@@ -29,67 +32,73 @@ class Configuration:
def save(self, filename: str) -> None:
"""Save configuration to zip archive"""
zip_stream = io.BytesIO()
with zipfile.ZipFile(zip_stream, "w") as zip_file:
zip_file.writestr(
self.ADJUSTMENT_FILE_NAME, json.dumps(
dataclasses.asdict(
self.field_adjustment)))
self.ADJUSTMENT_FILE_NAME,
json.dumps(dataclasses.asdict(self.field_adjustment)),
)
counter = 0
extension = ".png"
for square, card in self.catalogue:
counter += 1
file_stream = io.BytesIO()
np.save(
file_stream,
card_finder.simplify(square)[0],
allow_pickle=False)
fd, myfile = tempfile.mkstemp()
cv2.imwrite(myfile + extension, square)
file_name = ""
if isinstance(card, board.SpecialCard):
file_name = f's{card.value}-{card.name}-{counter}.npy'
file_name = f"s{card.value}-{card.name}-{counter}{extension}"
elif isinstance(card, board.NumberCard):
file_name = f'n{card.suit.value}{card.number}'\
f'-{card.suit.name}-{counter}.npy'
file_name = (
f"n{card.suit.value}{card.number}"
f"-{card.suit.name}-{counter}{extension}"
)
else:
raise AssertionError()
zip_file.writestr(
self.TEMPLATES_DIRECTORY + f"/{file_name}",
file_stream.getvalue())
zip_file.write(myfile + extension, arcname=f"{self.TEMPLATES_DIRECTORY}/{file_name}")
with open(filename, 'wb') as zip_archive:
with open(filename, "wb") as zip_archive:
zip_archive.write(zip_stream.getvalue())
@staticmethod
def load(filename: str) -> 'Configuration':
def load(filename: str) -> "Configuration":
"""Load configuration from zip archive"""
def _parse_file_name(card_filename: str) -> board.Card:
assert card_filename.startswith(
Configuration.TEMPLATES_DIRECTORY + '/')
pure_name = card_filename[
len(Configuration.TEMPLATES_DIRECTORY + '/'):]
if pure_name[0] == 's':
assert card_filename.startswith(Configuration.TEMPLATES_DIRECTORY + "/")
pure_name = card_filename[len(Configuration.TEMPLATES_DIRECTORY + "/") :]
if pure_name[0] == "s":
return board.SpecialCard(int(pure_name[1]))
if pure_name[0] == 'n':
if pure_name[0] == "n":
return board.NumberCard(
suit=board.NumberCard.Suit(
int(pure_name[1])), number=int(pure_name[2]))
suit=board.NumberCard.Suit(int(pure_name[1])),
number=int(pure_name[2]),
)
raise AssertionError()
catalogue: List[Tuple[np.ndarray, board.Card]] = []
with zipfile.ZipFile(filename, 'r') as zip_file:
with zipfile.ZipFile(filename, "r") as zip_file:
adj = adjustment.Adjustment(
**json.loads(
zip_file.read(Configuration.ADJUSTMENT_FILE_NAME)))
**json.loads(zip_file.read(Configuration.ADJUSTMENT_FILE_NAME))
)
mydir=tempfile.mkdtemp()
for template_filename in (
x for x in zip_file.namelist() if
x.startswith(Configuration.TEMPLATES_DIRECTORY + '/')):
x
for x in zip_file.namelist()
if x.startswith(Configuration.TEMPLATES_DIRECTORY + "/")
):
myfile = zip_file.extract(template_filename, path=mydir)
catalogue.append(
(np.load(io.BytesIO(zip_file.read(template_filename))),
_parse_file_name(template_filename)))
(
cv2.imread(myfile),
_parse_file_name(template_filename),
)
)
assert catalogue[-1][0] is not None
return Configuration(adj=adj, catalogue=catalogue, meta={})
@staticmethod
def generate(image: np.ndarray) -> 'Configuration':
def generate(image: np.ndarray) -> "Configuration":
"""Generate a configuration with user input"""
adj = adjustment.adjust_field(image)
squares = card_finder.get_field_squares(image, adj, 5, 8)