带线表格据gt生成无线表格

json解析练习，python图像处理练习，表格包围框毛刺问题待解决。

 # -*- coding: utf-8 -*-
 # coding: utf-8
 from PIL import Image, ImageDraw
 import cv2
 import os
 import csv
 import json
 color = {
     0: [255, 0, 0],
     1: [0, 255, 0],
     2: [0, 0, 255]
 }
 from PIL import Image
 import numpy as np

 # 取批量点取RGB众数，来推算背景色值
 def publicnum(num, d=0):
     dictnum = {}
     for i in range(len(num)):
         if str(num[i]) in dictnum.keys():
             dictnum[str(num[i])] += 1
         else:
             dictnum.setdefault(str(num[i]), 1)
     maxnum = 0
     maxkey = '[255 255 255]'
     for k, v in dictnum.items():
         if v >= maxnum:
             maxnum = v
             maxkey = k
     return maxkey

 page_img_dir = "JPG"
 output_dir = "年报_PDF_TABLE_JPG_eliminate_lines-5-15-final"

 isExists = os.path.exists(output_dir)
 if not isExists:
     os.makedirs(output_dir)
 csv_file = csv.reader(open('表格结构标注-带线年报_UTF-8.csv', 'r', encoding="gbk"))
 data_list = []
 for data in csv_file:
     data_list.append(data)

 print("page number: ", len(data_list) - 1)

 for data in data_list[1:]:  # 跳过第一行
     img_path = data[0]
     img_name = img_path.split('/')[-1]  # /分割后最后一个为名字
     pdf_name = img_name.split('_')[0]  # -分割后 第一个是名字
     local_img_path = os.path.join(page_img_dir, img_name)  # 拼接路径
     print(local_img_path)
     annotation = json.loads(data[2])  # json单元格读取
     objects = annotation['objects']  # object是一个列表，读取该列表
     cnt = 0
     tu = Image.open(local_img_path)
     page_img = np.array(tu)
     for page_object in objects:
         if 'cur' in page_object.keys():
             cur = page_object['cur']
         else:
             cur = cnt
         polygon = page_object['polygon']['ptList']
         x_list = [p['x'] for p in polygon]
         y_list = [p['y'] for p in polygon]
         x_min = min(x_list)
         x_max = max(x_list)
         y_min = min(y_list)
         y_max = max(y_list)
         if abs(x_max - x_min) < 20:  # 纵向线条
             xx = int((x_min + x_max) / 2)
             inline_y_list = [y_max+20, y_min]
             #寻找相交横线分割点
             for in_page_object in objects:
                 in_polygon = in_page_object['polygon']['ptList']
                 in_x_list = [in_p['x'] for in_p in in_polygon]
                 in_y_list = [in_p['y'] for in_p in in_polygon]
                 in_x_min = min(in_x_list)
                 in_x_max = max(in_x_list)
                 in_y_min = min(in_y_list)
                 in_y_max = max(in_y_list)
                 if in_y_max - in_y_min < 20:  # 判断为横线
                     if in_x_max+5 >= xx and in_x_min-5 <= xx:  # 判断相交
                         point_y = in_y_min
                         inline_y_list.append(point_y)
                         if 0<abs(y_max-point_y)< 10:
                             try:
                                 inline_y_list.remove(max(y_max+20, point_y))
                                 inline_y_list.append(min(y_max+20, point_y))
                             except:
                                 pass
                         elif 0 < abs(y_min-point_y) < 10:
                             try:
                                 inline_y_list.remove(min(y_min, point_y))
                                 inline_y_list.append(max(y_min, point_y))
                             except:
                                 pass
             inline_y_list = list({}.fromkeys(inline_y_list).keys())
             inline_y_list.sort()
             inline_y_list[-1]+=5
             if inline_y_list[-1]>2339:
                 inline_y_list[-1]=2339
             # 线条分割结束
             for i in range(0, inline_y_list.__len__()):
                 if i < inline_y_list.__len__() - 1:
                     # 开始取样
                     back_colors = []
                     for yy in range(inline_y_list[i], inline_y_list[i + 1]):
                         if xx + 8 < 1654:
                             back_colors.append(page_img[yy, xx + 8])
                         else:
                             back_colors.append(page_img[yy, xx - 8])
                     back_color = publicnum(back_colors)
                     back_color = back_color[1:-1]
                     try:
                         back_color = back_color.split(' ')
                     except:
                         back_color = back_color.split('   ')
                         print(type(back_color))
                     if len(back_color) > 3:
                         back_color = list(filter(None, back_color))
                     # 取样结束
                     # 纵向填色

                     for yy in range(inline_y_list[i]-4, inline_y_list[i + 1]-4):
                         if y_min-20<inline_y_list[i]<y_max+20 :
                             for ranging in range(-4, x_max-x_min+5):
                                 if x_min+ranging < 1654 and x_min+ranging >= 0:
                                         page_img[yy, x_min + ranging] = back_color
                                 else:
                                     pass
                         else:
                             pass
         elif abs(y_max - y_min) < 20:  # 横向线条
             yy = int((y_min + y_max) / 2)
             inline_x_list = [x_max+20, x_min]
             # 寻找相交横线分割点
             for in_page_object2 in objects:
                 polygon2 = in_page_object2['polygon']['ptList']
                 in_x_list2 = [p['x'] for p in polygon2]
                 in_y_list2 = [p['y'] for p in polygon2]
                 in_x_min = min(in_x_list2)
                 in_x_max = max(in_x_list2)
                 in_y_min = min(in_y_list2)
                 in_y_max = max(in_y_list2)
                 if abs(in_x_max - in_x_min) < 20:  # 判断为纵线
                     if in_y_max+5 >= y_min and in_y_min-5 <= y_max:  # 判断相交
                         point_x = in_x_min
                         inline_x_list.append(point_x)
                         if 0<abs(x_max-point_x)<10:
                             try:
                                 inline_x_list.remove(max(x_max+20, point_x))
                                 inline_x_list.append(min(x_max+20, point_x))
                             except:
                                 pass
                         elif 0<abs(x_min-point_x)<10:
                             try:
                                 inline_x_list.remove(min(x_min, point_x))
                                 inline_x_list.append(max(x_min, point_x))
                             except:
                                 pass
                         else:
                             pass
             inline_x_list = list({}.fromkeys(inline_x_list).keys())
             inline_x_list.sort()
             #inline_x_list[-1]+=5
             # 线条分割结束
             for i in range(0, inline_x_list.__len__()):
                 if i < inline_x_list.__len__() - 1:
                     # 开始取样
                     back_colors = []
                     for xx in range(inline_x_list[i], inline_x_list[i + 1]):
                         if yy+8 < 2339:
                             back_colors.append(page_img[yy + 8, xx])
                         else:
                             back_colors.append(page_img[yy - 8, xx])
                     back_color = publicnum(back_colors)
                     back_color = back_color[1:-1]
                     try:
                         back_color = back_color.split(' ')
                     except:
                         back_color = back_color.split('   ')

                     if len(back_color) > 3:
                         back_color = list(filter(None, back_color))
                     # 取样结束
                     # 横线填色
                     for xx in range(inline_x_list[i]-4, inline_x_list[i + 1]-4):
                         if x_min-20<inline_x_list[i]<x_max+20:
                             for ranging in range(-4, y_max-y_min+5):
                                 if y_min+ranging < 2339 and y_min+ranging >= 0:
                                     page_img[y_min+ranging, xx] = back_color
                                 elif y_min+ranging>=2339:
                                     page_img[2338, xx] = back_color
                                 else:
                                     page_img[0, xx] = back_color
                         else:
                             pass
         else:
             print("no such line", 'x_min:', x_min,'x_max:', x_max, 'y_max:', y_max, 'y_min:', y_min)
     tu = Image.fromarray(page_img.astype('uint8'))
     output_path = os.path.join(output_dir, img_name.split('.')[0] + '_' + str(cur) + ".jpg")
     tu.save(output_path)
     cv2.imwrite(output_path, page_img)
     cnt += 1