如何改进OpenCV代码实现图像字符数字按列分类分文件夹有序保存

我已通过以下代码将每个字母和数字裁剪出来，效果如下：

使用的代码：

import cv2
import numpy as np
import glob
from google.colab.patches import cv2_imshow
img = cv2.imread('/content/drive/MyDrive/Dataset/S__23363591.jpg')
image = cv2.resize(img,None,None,0.4,0.4)
#cv2_imshow(image)

gray = cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)
ret,thresh = cv2.threshold(gray,190,255,cv2.THRESH_BINARY_INV)
kernel = np.ones((1,1), np.uint8)
img_dilation = cv2.dilate(thresh, kernel, iterations=1)
ctrs,_= cv2.findContours(img_dilation.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
sorted_ctrs = sorted(ctrs, key=lambda ctr: cv2.boundingRect(ctr)[0])
for i, ctr in enumerate(sorted_ctrs):
    x, y, w, h = cv2.boundingRect(ctr)
    roi = image[y:y+h, x:x+w]
    cv2.rectangle(image,(x,y),( x + w, y + h ),(36,255,12),2)
    if w >2  and h > 2:
      #os.chdir('{}'.format(folder_out))
      roi = cv2.resize(roi,(224,224))
      #cv2.imwrite('letter{}.jpg'.format(i), roi)
cv2_imshow(image)

我希望实现以下功能：

• 将同一列的图像保存到同一个文件夹，效果如下：
  
• 每个列文件夹下再分别创建字母和数字子文件夹，对应保存两类图像，效果如下：
  
• 图像按顺序命名，效果如下：
  

请提供能实现上述图像按列分类分文件夹、有序命名功能的改进代码。

改进后的代码 #

import cv2
import numpy as np
import os
from google.colab.patches import cv2_imshow
import pytesseract

# Colab环境需先执行安装命令
# !pip install pytesseract
# !sudo apt install tesseract-ocr

# 读取并缩放图像
img = cv2.imread('/content/drive/MyDrive/Dataset/S__23363591.jpg')
image = cv2.resize(img, None, None, 0.4, 0.4)

# 图像预处理
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
ret, thresh = cv2.threshold(gray, 190, 255, cv2.THRESH_BINARY_INV)
kernel = np.ones((1, 1), np.uint8)
img_dilation = cv2.dilate(thresh, kernel, iterations=1)

# 查找轮廓并按X坐标排序
ctrs, _ = cv2.findContours(img_dilation.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
sorted_ctrs = sorted(ctrs, key=lambda ctr: cv2.boundingRect(ctr)[0])

# 按列分组轮廓（X坐标差值阈值可根据实际调整）
column_groups = []
current_group = []
prev_x = None
x_threshold = 20

for ctr in sorted_ctrs:
    x, _, _, _ = cv2.boundingRect(ctr)
    if prev_x is None or abs(x - prev_x) <= x_threshold:
        current_group.append(ctr)
    else:
        column_groups.append(current_group)
        current_group = [ctr]
    prev_x = x
if current_group:
    column_groups.append(current_group)

# 创建根保存目录
root_save_dir = '/content/drive/MyDrive/Dataset/char_columns'
os.makedirs(root_save_dir, exist_ok=True)

# 处理每一列字符
for col_idx, col_ctrs in enumerate(column_groups, start=1):
    # 创建列文件夹及子文件夹
    col_dir = os.path.join(root_save_dir, f'column_{col_idx}')
    letters_dir = os.path.join(col_dir, 'letters')
    digits_dir = os.path.join(col_dir, 'digits')
    os.makedirs(col_dir, exist_ok=True)
    os.makedirs(letters_dir, exist_ok=True)
    os.makedirs(digits_dir, exist_ok=True)
    
    # 按Y坐标排序（从上到下）
    sorted_col_ctrs = sorted(col_ctrs, key=lambda ctr: cv2.boundingRect(ctr)[1])
    
    # 逐个处理字符
    for char_idx, ctr in enumerate(sorted_col_ctrs, start=1):
        x, y, w, h = cv2.boundingRect(ctr)
        if w > 2 and h > 2:
            roi = image[y:y+h, x:x+w]
            roi_resized = cv2.resize(roi, (224, 224))
            
            # 识别字符并分类
            roi_gray = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
            _, roi_thresh = cv2.threshold(roi_gray, 190, 255, cv2.THRESH_BINARY_INV)
            char_text = pytesseract.image_to_string(roi_thresh, config='--psm 10 --oem 3 -c tessedit_char_whitelist=ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789').strip()
            
            # 保存到对应文件夹
            if char_text.isdigit():
                save_path = os.path.join(digits_dir, f'{char_idx}.jpg')
            elif char_text.isalpha():
                save_path = os.path.join(letters_dir, f'{char_idx}.jpg')
            else:
                continue  # 无法识别的字符跳过
            
            cv2.imwrite(save_path, roi_resized)
            cv2.rectangle(image, (x, y), (x + w, y + h), (36, 255, 12), 2)

cv2_imshow(image)

代码说明 #

1. 列分组逻辑：通过判断轮廓X坐标的差值，将同一列的字符归为一组，阈值x_threshold可根据图像实际间距调整
2. 字符分类：使用Tesseract OCR识别字符，通过白名单限定识别范围，区分字母和数字后分别保存
3. 文件夹结构：自动创建column_N列文件夹，每个列文件夹下生成letters和digits子文件夹
4. 有序命名：每列内的字符按从上到下顺序命名为1.jpg、2.jpg等

内容的提问来源于stack exchange，提问作者Nara Nart