如何用OpenCV去除图像中的线条且不损坏其余内容？

我有一张带有字符和水平线条的图像，需要去除线条但完整保留所有字符。

首次尝试及问题 #

我先用以下代码尝试去除水平线条：

image = cv2.imread('im.png')
gray = cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

# Remove horizontal lines
horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (20,1))
remove_horizontal = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, horizontal_kernel, iterations=2)

cnts = cv2.findContours(remove_horizontal, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
for c in cnts:
    cv2.drawContours(result, [c], -1, (255,255,255), 5)
plt.imshow(result)

处理后发现，和线条相连的字符部分也被一并删除了，这些字符是必须保留的。

二次尝试及残留问题 #

我又尝试添加修复逻辑，代码如下：

image = cv2.imread('connected/im.png')
gray = cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

# Remove horizontal
horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (40,1))
detected_lines = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, horizontal_kernel, iterations=2)
cnts = cv2.findContours(detected_lines, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
for c in cnts:
    cv2.drawContours(image, [c], -1, (255,255,255), 2)

# Repair image
repair_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3,6))
result = 255 - cv2.morphologyEx(255 - image, cv2.MORPH_CLOSE, repair_kernel, iterations=2)

plt.imshow(result)

虽然修复了部分字符，但图像出现明显颗粒感，仍能看到线条残留的痕迹。

请问有没有更好的图像修复/线条去除方法？

这里提供几种更精准的线条去除+字符修复方法，针对你的场景可以按需选择：

方法1：结合形态学检测与图像修复（Inpaint） #

先精准定位线条区域，再用图像修复算法填充线条位置，参考周围像素还原字符：

import cv2
import numpy as np

# 读取图像并预处理
image = cv2.imread('im.png')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

# 检测水平线条：用长核确保只捕获线条，不包含字符
horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (50, 1))
detected_lines = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, horizontal_kernel, iterations=1)

# 生成修复掩码：标记需要修复的线条区域
mask = np.zeros(image.shape[:2], np.uint8)
cnts = cv2.findContours(detected_lines, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
for c in cnts:
    # 控制线条绘制宽度，避免覆盖过多字符
    cv2.drawContours(mask, [c], -1, 255, 2)

# 使用Telea算法修复图像，适配纹理/结构类修复场景
result = cv2.inpaint(image, mask, 3, cv2.INPAINT_TELEA)

cv2.imshow('Result', result)
cv2.waitKey(0)

优势：精准区分线条与字符，修复时参考周边像素填充，不会破坏字符完整性，无颗粒感残留。

方法2：改进形态学流程，分离线条与字符 #

通过膨胀-分离-腐蚀的操作顺序，先让字符与线条断开连接，再去除线条：

import cv2
import numpy as np

image = cv2.imread('im.png')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

# 先膨胀字符，让粘连的字符与线条分离
dilate_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (2, 2))
dilated = cv2.dilate(thresh, dilate_kernel, iterations=1)

# 检测线条，此时线条与字符已无粘连
horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (40, 1))
detected_lines = cv2.morphologyEx(dilated, cv2.MORPH_OPEN, horizontal_kernel, iterations=1)

# 从膨胀后的图像中减去线条区域，再腐蚀回字符原始大小
result = cv2.subtract(dilated, detected_lines)
erode_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (2, 2))
result = cv2.erode(result, erode_kernel, iterations=1)

# 转RGB格式显示
result = cv2.cvtColor(result, cv2.COLOR_GRAY2BGR)
cv2.imshow('Result', result)
cv2.waitKey(0)

优势：避免线条与字符粘连导致的误删，修复后的字符轮廓完整，无残留痕迹。

方法3：基于轮廓特征过滤，只删除线条 #

线条的轮廓通常是细长型，与字符的轮廓面积、宽高比差异明显，可通过轮廓特征区分：

import cv2
import numpy as np

image = cv2.imread('im.png')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

# 提取所有轮廓
cnts = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]

# 创建掩码，保留字符轮廓，过滤线条轮廓
mask = np.zeros(thresh.shape, np.uint8)
for c in cnts:
    area = cv2.contourArea(c)
    x, y, w, h = cv2.boundingRect(c)
    # 线条的宽高比远大于字符，且面积相对均匀
    if w / h > 20 and area > 100:
        continue
    cv2.drawContours(mask, [c], -1, 255, -1)

# 用掩码提取字符，背景填充白色
result = np.full(image.shape, 255, dtype=np.uint8)
result[mask == 255] = image[mask == 255]

cv2.imshow('Result', result)
cv2.waitKey(0)

优势：完全基于轮廓特征区分线条和字符，适合线条与字符形态差异明显的场景，处理效果干净彻底。

内容的提问来源于stack exchange，提问作者InterestingQuestions44