From c082b525b5e501dfa24038e3a85e444d88bfb1d0 Mon Sep 17 00:00:00 2001
From: admin <weikou2014>
Date: 星期五, 28 四月 2023 19:33:15 +0800
Subject: [PATCH] 开盘啦采集工具/看盘页面优化
---
opencv_util.py | 240 ++++++++++++++++++++++++++++++++++++++++++++++++++++--------
1 files changed, 208 insertions(+), 32 deletions(-)
diff --git a/opencv_util.py b/opencv_util.py
index d99fc5b..c87fe45 100644
--- a/opencv_util.py
+++ b/opencv_util.py
@@ -1,69 +1,245 @@
# 浜屽�煎寲鍥惧儚
+import os
+import random
+
import cv2
+import matplotlib.pyplot as plt
+import numpy
+
+SHOW_PLT = False
def gray_img(img):
+ if img.ndim == 2:
+ return img
result = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
return result
-def is_col_empty(img, col, thresh_hold=0):
- for r in img:
- if r[col] > thresh_hold:
+def is_col_empty(img, col, start_row, end_row, thresh_hold=0):
+ for r in range(start_row, end_row + 1):
+ if img[r][col] > thresh_hold:
return False
return True
-def is_col_full(img, col, thresh_hold_start=64, thresh_hold_end=255):
+def is_col_full(img, col, start_row, end_row, thresh_hold_start=64, thresh_hold_end=255, thresh_empty_count=1):
rows, cols = img.shape
- for r in img:
- if r[col] < thresh_hold_start or r[col] > thresh_hold_end:
+ empty_count = 0
+ for r in range(start_row, end_row + 1):
+ if img[r][col] < thresh_hold_start or img[r][col] > thresh_hold_end:
+ empty_count += 1
+ if empty_count > thresh_empty_count:
+ return False
+ return True
+
+
+# 琛屾槸鍚︽槸婊＄殑
+def is_row_full(img, row, start_col, end_col, thresh_hold_start=64, thresh_hold_end=255):
+ rows, cols = img.shape
+ for c in range(start_col, end_col + 1):
+ if img[row][c] < thresh_hold_start or img[row][c] > thresh_hold_end:
return False
return True
+
+
+def format_num_img(img):
+ # 鍏堣皟鏁翠负鏂瑰舰
+ rows, cols = img.shape
+ img_new = numpy.zeros((max(rows, cols), max(rows, cols)), numpy.uint8)
+ img_new.fill(0)
+ start_col = (max(rows, cols) - cols) // 2
+ if rows > cols:
+ for r in range(rows):
+ for c in range(cols):
+ img_new[r][c + start_col] = img[r][c]
+ else:
+ pass
+ # 璁剧疆涓�28X28鍍忕礌
+ return cv2.resize(img_new, (28, 28), interpolation=cv2.INTER_AREA)
+
+
+# 鍒嗛殧鏁板瓧
+def split_nums_img(img):
+ # 鑾峰彇浠ｇ爜鐨勬渶澶ц壊鍊�
+ rows, cols = img.shape
+ max_color = 0
+ for r in range(0, rows):
+ for c in range(0, cols):
+ if img[r][c] > max_color:
+ max_color = img[r][c]
+
+ start_index = -1
+ end_index = -1
+ codes_pos = []
+ for col in range(0, cols):
+ if not is_col_empty(img, col, 0, rows - 1, max_color * 2 // 3):
+ if start_index < 0:
+ start_index = col
+ end_index = col
+ else:
+ if end_index >= 0 and start_index >= 0:
+ codes_pos.append((start_index, end_index + 1))
+ end_index = -1
+ start_index = -1
+ img_detail = []
+ for i in range(len(codes_pos)):
+ temp_img = img[0:rows - 1, codes_pos[i][0]:codes_pos[i][1]]
+ temp_img = format_num_img(temp_img)
+ img_detail.append(temp_img)
+ return img_detail
# 鍒嗙鍚岃姳椤虹殑浠ｇ爜
def clip_ths_code_area(img):
img = gray_img(img)
rows, cols = img.shape
- end_col = cols - 1
- for col in range(0, cols, 1):
- if is_col_full(img, col, 38, 38):
+
+ # 琛屽垎闅�
+ full_row = -1
+ start_row = -1
+ end_row = -1
+ for r in range(0, rows, 1):
+ if is_row_full(img, r, 0, cols // 2, 38, 38):
# print("鎵惧埌鍒嗛殧锛�", col)
- end_col = col - 1
+ full_row = r
+ if start_row >= 0 and r - start_row > 10:
+ end_row = r - 1
+ break
+
+ else:
+ if full_row > -1:
+ full_row = -1
+ if start_row < 0:
+ start_row = r
+ if end_row < 0:
+ end_row = rows - 1
+
+ if start_row < 0:
+ raise Exception("娌℃壘鍒颁笂鍒嗗壊绾�")
+
+ end_col = cols - 1
+ for c in range(cols - 1, -1, -1):
+ if is_col_full(img, c, start_row, end_row, 38, 38, 2):
+ # print("鎵惧埌鍒嗛殧锛�", col)
+ end_col = c - 1
break
# 寰�鍓嶆壘鏁板瓧鍒嗛殧
content_start = -1
empty_start = -1
empty_end = -1
- start_index = 0
- end_index = end_col
+ start_index = -1
+ end_index = -1
+ codes_pos = []
+ # 鑾峰彇浠ｇ爜鐨勬渶澶ц壊鍊�
+ max_color = 0
+ for r in range(start_row, end_row + 1):
+ for c in range(end_col, end_col // 2, -1):
+ if img[r][c] > max_color:
+ max_color = img[r][c]
+
for col in range(end_col, 0, -1):
- if not is_col_empty(img, col):
- if content_start < 0:
- content_start = col
- empty_start = -1
- empty_end = -1
+ if not is_col_empty(img, col, start_row, end_row, max_color * 2 // 3):
+ if start_index < 0:
+ start_index = col
+ end_index = col
else:
- if empty_start < 0:
- empty_start = col
- empty_end = col
- else:
- empty_end = col
- if empty_start - empty_end > 2 and content_start > 0:
- start_index = col + (empty_start - empty_end + 1)
- end_index = content_start
- # print("浠ｇ爜鑼冨洿锛�", start_index, end_index)
- break
- clip_img = img[0:rows, start_index:end_index]
+ if end_index >= 0 and start_index >= 0:
+ codes_pos.append((end_index - 1, start_index + 1))
+ end_index = -1
+ start_index = -1
+ codes_pos = codes_pos[:6]
+ codes_pos.sort(key=lambda x: x[0])
+
+ if SHOW_PLT:
+ plt.figure(figsize=(10, 4))
+ img_detail = []
+ for i in range(len(codes_pos)):
+ if SHOW_PLT:
+ plt.subplot(2, 5, i + 1)
+ plt.title(f'pred {i}')
+ plt.axis('off')
+ temp_img = img[start_row:end_row, codes_pos[i][0]:codes_pos[i][1]]
+ temp_img = format_num_img(temp_img)
+ img_detail.append(temp_img)
+ if SHOW_PLT:
+ plt.imshow(temp_img, cmap='gray')
+ if SHOW_PLT:
+ plt.show()
+
+ clip_img = img[start_row:end_row, codes_pos[0][0]:codes_pos[-1][1]]
# cv2.imwrite("test1.png", clip_img)
- return clip_img
+ return clip_img, img_detail
# print(clip_img.shape)
# ret1, p1 = cv2.threshold(src=clip_img, thresh=100, maxval=255, type=cv2.THRESH_BINARY)
# cv2.imwrite("D:/workspace/GP/trade_desk/test3.png", p1)
-if __name__ == "__main__":
- img = gray_img(cv2.imread("C:\\Users\\Administrator\\Desktop\\test.png"))
- cv2.imwrite("C:\\Users\\Administrator\\Desktop\\test_gray.png", img)
+def __test4():
+ files = os.listdir("datas/test4/")
+ for file in files:
+ code = file[:-4]
+ img = cv2.imread(f"datas/test4/{file}", cv2.IMREAD_GRAYSCALE)
+ img_details = split_nums_img(img)
+ for d in range(0, len(img_details)):
+ cv2.imwrite(f"C:/Users/Administrator/Desktop/ocr/codes/{code}_{random.randint(0, 100000)}.png",
+ img_details[d])
+ plt.figure(figsize=(10, 4))
+ for i in range(0, len(img_details)):
+ plt.subplot(2, 5, i + 1)
+ plt.title(f'pred {i}')
+ plt.axis('off')
+ plt.imshow(img_details[i], cmap='gray')
+ plt.show()
+
+
+def __test3():
+ files = os.listdir("datas/test3/")
+ for file in files:
+ code = file[:-4]
+ img = cv2.imread(f"datas/test3/{file}", cv2.IMREAD_GRAYSCALE)
+ rows, cols = img.shape
+ for r in range(rows):
+ for c in range(cols):
+ img[r][c] = 255 - img[r][c]
+
+ img_details = split_nums_img(img)
+ for d in range(0, len(img_details)):
+ cv2.imwrite(f"C:/Users/Administrator/Desktop/ocr/codes/{code}_{random.randint(0, 100000)}.png",
+ img_details[d])
+ plt.figure(figsize=(10, 4))
+ for i in range(0, len(img_details)):
+ plt.subplot(2, 5, i + 1)
+ plt.title(f'pred {i}')
+ plt.axis('off')
+ plt.imshow(img_details[i], cmap='gray')
+ plt.show()
+
+
+if __name__ == '__main__':
+ __test4()
+
+if __name__ == "__main__1":
+ #
+ files = os.listdir("datas/test/")
+ for file in files:
+ code = file[:6]
+ img = cv2.imread(f"datas/test/{file}", cv2.IMREAD_GRAYSCALE)
+ img, img_details = clip_ths_code_area(img)
+ for d in range(0, len(img_details)):
+ cv2.imwrite(f"C:/Users/Administrator/Desktop/ocr/codes/{code}_{random.randint(0, 100000)}.png",
+ img_details[d])
+ if SHOW_PLT:
+ plt.figure(figsize=(1, 1))
+ plt.subplot(1, 1, 1)
+ plt.title(f"test")
+ plt.axis('off')
+ plt.imshow(img)
+ plt.show()
+ pass
+ # img = gray_img(cv2.imread("C:\\Users\\Administrator\\Desktop\\ocr\\code_test.png"))
+ # h = img.shape[0]
+ # w = img.shape[1]
+ # img = cv2.resize(img, (int(w * 2), int(h * 2)), interpolation=cv2.INTER_AREA)
+ # cv2.imwrite("C:\\Users\\Administrator\\Desktop\\ocr\\code_test_gray.png", img)
--
Gitblit v1.8.0