使用OpenCV findContours提取工程图纸表格结果异常的问题求助

问题排查 #

你的原代码存在2处核心逻辑错误：

• 首次轮廓检测后将所有四边形轮廓填充为白色，直接清除了表格的边框特征，导致后续无法识别到有效表格区域
• 形态学操作误用了cv2.MORPH_OPEN，应该使用闭运算来连接断裂的表格线，且原核尺寸参数不匹配当前图纸的表格宽高比例

修复后可运行代码 #

import cv2
import numpy as np

image = cv2.imread('01.png')
original = image.copy()
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (3,3), 0)
thresh = cv2.threshold(blur, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

# 分别提取水平和垂直直线，组合成表格边框
# 提取水平直线
horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (40,1))
detect_horizontal = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, horizontal_kernel, iterations=2)
cnts_horizontal = cv2.findContours(detect_horizontal, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts_horizontal = cnts_horizontal[0] if len(cnts_horizontal) == 2 else cnts_horizontal[1]
for c in cnts_horizontal:
    cv2.drawContours(thresh, [c], -1, 255, 2)

# 提取垂直直线
vertical_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (1,40))
detect_vertical = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, vertical_kernel, iterations=2)
cnts_vertical = cv2.findContours(detect_vertical, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts_vertical = cnts_vertical[0] if len(cnts_vertical) == 2 else cnts_vertical[1]
for c in cnts_vertical:
    cv2.drawContours(thresh, [c], -1, 255, 2)

# 闭运算连接表格边框，形成完整的外框
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (10,10))
close = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel, iterations=3)

# 提取外轮廓，过滤非表格区域
cnts = cv2.findContours(close, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
# 按面积倒序，取最大的四边形就是目标表格
cnts = sorted(cnts, key=cv2.contourArea, reverse=True)
for c in cnts:
    peri = cv2.arcLength(c, True)
    approx = cv2.approxPolyDP(c, 0.02 * peri, True)
    if len(approx) == 4:
        x,y,w,h = cv2.boundingRect(approx)
        # 增加面积和宽高比过滤，避免识别到小矩形
        if cv2.contourArea(c) > 10000 and 1 < w/h < 5:
            cv2.drawContours(image, [approx], -1, (36,255,12), 3)
            crop = original[y:y+h, x:x+w]
            break

cv2.imwrite('image.png', image)
cv2.imwrite('ROI.png', crop)
cv2.waitKey()
cv2.destroyAllWindows()

优化点说明 #

• 新增了水平/垂直直线单独提取逻辑，过滤掉图纸中无关的文字、线条干扰
• 降低了轮廓拟合的epsilon参数从0.05到0.02，避免漏识别边缘有轻微变形的表格外框
• 增加了轮廓面积、宽高比过滤规则，以及按面积倒序取最大四边形的逻辑，适配工程图纸中目标表格为最大矩形的特征
• 调整了形态学运算类型和参数，保证表格边框的完整性

内容的提问来源于stack exchange，提问作者Fayne