[Python] 纯文本查看 复制代码
import time
from ctypes import windll
import numpy as np
import cv2
from mss import mss
class Mxbx:
def __init__(self):
print('欢迎使用')
windll.user32.SetProcessDPIAware()
self.sct = mss()
def GetCapture(self, x1, y1, x2, y2):
'''
截图
:param x1:
:param y1:
:param x2:
:param y2:
:return: rgb图像
'''
monitor = {'top': y1, 'left': x1, 'width': x2 - x1, 'height': y2 - y1}
pic = self.sct.grab(monitor)
pic = cv2.cvtColor(np.array(pic), cv2.COLOR_RGBA2RGB)
return pic
def parse_custom_color(self, color_string):
r, g, b = int(color_string[4:6], 16), int(color_string[2:4], 16), int(color_string[0:2], 16)
if len(color_string) > 6:
dr, dg, db = int(color_string[7:9], 16), int(color_string[9:11], 16), int(color_string[11:13], 16)
else:
dr, dg, db = 0, 0, 0
lower_rgb = np.array([max(0, r - dr), max(0, g - dg), max(0, b - db)], dtype=np.uint8)
upper_rgb = np.array([min(255, r + dr), min(255, g + dg), min(255, b + db)], dtype=np.uint8)
return lower_rgb, upper_rgb
def Stressshow(self, x1, y1, x2, y2, colors, types=0):
'''
隐藏其他颜色为黑色，只保留选中颜色
:param x1: 起点x
:param y1: 起点Y
:param x2: 终点x
:param y2: 终点Y
:param colors: 十六进制颜色，可以从大漠的色彩描述直接获取
:param types: 0：原来颜色，非零：白色
:return:
'''
image = self.GetCapture(x1, y1, x2, y2)
colors = colors.split("|")
masks = np.zeros_like(image[:, :, 0], dtype=np.uint8)
for color in colors:
lower_rgb, upper_rgb = self.parse_custom_color(color)
mask = cv2.inRange(image, lower_rgb, upper_rgb)
masks += mask
result = cv2.bitwise_and(image, image, mask=masks)
if types:
result[np.where(masks != 0)] = [255, 255, 255]
return result
def SetDict(self, lines):
'''
初始化字库文件
:param lines: 字库数组
:return: 点阵数据
'''
des = {}
for line in lines[1:]:
parts = line.split("$")
hex_str = parts[0]
hex_values = np.array([int(c, 16) for c in hex_str], dtype=np.uint8)
bin_values = np.unpackbits(hex_values[:, None], axis=1).reshape(-1, 8)[:, 4:] # 只需要后4位
num_elements = bin_values.shape[0]
remainder = num_elements % 11
if remainder != 0:
padding = 11 - remainder
bin_values = np.pad(bin_values, ((0, padding), (0, 0)), mode='constant', constant_values=0)
bin_values = bin_values.reshape(-1, 11).T
des.setdefault(parts[1], bin_values)
return lines[0], des
def Ocr(self, x1, y1, x2, y2, thd, DIict):
'''
点阵文字识别
:param x1:
:param y1:
:param x2:
:param y2:
:param thd: 相似度
:param DIict: 字库
:return:
'''
def filter_points(points, max_distance):
if points.shape[0] <= 1:
return points
diff = points[:, None, :] - points[None, :, :]
distances = np.linalg.norm(diff, axis=2)
mask = distances < max_distance
to_remove = np.zeros(len(points), dtype=bool)
for i in range(len(points)):
if to_remove[i]:
continue
for j in range(i + 1, len(points)):
if mask[i, j]:
to_remove[j] = True
cleaned_data = points[~to_remove]
return cleaned_data
img = self.Stressshow(x1, y1, x2, y2, DIict[0], 1)
img = (img != 0).any(axis=2).astype(int)
img = np.array(img, dtype=np.uint8)
res = {}
for key, value in DIict[1].items():
_, h = value.shape
if x2 - x1 > h and y2 - y1 > 11:
result = cv2.matchTemplate(value, img, cv2.TM_CCOEFF_NORMED)
loc = np.column_stack(np.where(result >= thd))
loc = filter_points(loc, min(6, h))
if loc.size > 1:
py, px = 6 + y1, int(h / 2 + x1)
loc += (py, px)
loc = loc[:, [1, 0]]
res.setdefault(key, loc)
return res if len(res) > 0 else 0
def Getocr(self, input_data, x_size=15):
'''
解析ocr
:param input_data:ocr返回值
:param x_size: x间距，距离小合并到一起
:return:
'''
output = [(key, tuple(row)) for key, arr in input_data.items() for row in arr]
output.sort(key=lambda x: x[1][1])
groups = []
current_group = []
for i, (char, coord) in enumerate(output):
if current_group:
last_coord = current_group[-1][1]
if abs(coord[1] - last_coord[1]) <= 5:
current_group.append((char, coord))
else:
groups.append(current_group)
current_group = [(char, coord)]
else:
current_group.append((char, coord))
if current_group:
groups.append(current_group)
def merge_by_x_coordinate(groups, size=x_size):
merged_groups = []
for group in groups:
group.sort(key=lambda x: x[1][0])
merged_group = []
current_subgroup = []
for i, (char, coord) in enumerate(group):
if not current_subgroup:
current_subgroup.append((char, coord))
else:
last_char, last_coord = current_subgroup[-1]
if abs(coord[0] - last_coord[0]) <= size:
current_subgroup.append((char, coord))
else:
merged_group.append(current_subgroup)
current_subgroup = [(char, coord)]
if current_subgroup:
merged_group.append(current_subgroup)
merged_groups.append(merged_group)
return merged_groups
merged_groups = merge_by_x_coordinate(groups, size=x_size)
def merge_characters(merged_groups):
merged_dict = {}
for group in merged_groups:
for subgroup in group:
merged_char = ''.join([char for char, _ in subgroup])
first_coord = subgroup[0][1]
if merged_char not in merged_dict:
merged_dict[merged_char] = []
merged_dict[merged_char].append(first_coord)
return merged_dict
merged_dict = merge_characters(merged_groups)
return merged_dict
a = Mxbx()
st = time.time()
# 初始化字库数据，第一行为颜色值，后续为字库数据
zk1 = a.SetDict(['b6ffdb-202020|ffffff-202020|ffb666-303030|90dbff-202020|ffffdb-202020',
'2080525220500D21A1FD44C0D0560A864008110$梁$0.0.52$13',
'04108610C604A4030$3$0.0.16$11',
'200001007FF$1$0.0.13$11',
'268C1A00C01803001C60$0$0.0.20$11',
])
# 这个字库数据是电脑右下角时间及日期
zk2 = a.SetDict(['d0d4d5-303030',
'200401007FFFFE$1$0.0.25$11',
'40F8360CC11863F85E08$2$0.0.32$11',
'40508610C21CE6F78060$3$0.0.28$11',
'0100E0340C8711FFFFF808$4$0.0.37$11',
'085F0620C418C70F80E0$5$0.0.28$11',
'1F8ED920C4188318E1F0$6$0.0.33$11',
'80100201C0F871383C0600$7$0.0.24$11',
'33CFCE10C21843FCDCF0$8$0.0.39$11',
'7C58C608C11826DF8FC0$9$0.0.35$11',
'3F8C1E00C01803C19FE0$0$0.0.31$11'])
b = a.Ocr(1763, 1032, 1855, 1079, 0.9, zk2)
if b:
b = a.Getocr(b, 15)
print(time.time() - st)
print(b)
[复制链接]
发表于 2025-1-9 14:06
|
发表于 2025-1-9 14:48
