如何用Python+OpenCV检测毛孔（圆形）及直径？代码检测异常求助

我正在开发一个使用OpenCV检测毛孔（圆形及其直径）的程序，目前程序可检测轮廓但无法识别毛孔，输入图片如下：

当前使用的代码如下：

import cv2
import numpy as np

image_paths = ["image_with_contours2.jpg"]

for i, image_path in enumerate(image_paths):
    # Read the image
    image = cv2.imread(image_path)

    # Convert the image to grayscale
    img_gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

    # Apply Gaussian blur to reduce noise
    img_blur = cv2.GaussianBlur(img_gray, (5, 5), 0)

    # Apply Hough Circle Transform to detect circles
    circles = cv2.HoughCircles(img_blur, cv2.HOUGH_GRADIENT, dp=1, minDist=50, param1=50, param2=30, minRadius=10, maxRadius=100)

    # Draw detected circles on the original image
    if circles is not None:
        circles = np.round(circles[0, :]).astype(int)
        for (x, y, radius) in circles:
            cv2.circle(image, (x, y), radius, (0, 0, 255), 2)

    # Save the image with detected circles
    cv2.imwrite(f'image_with_circles{i}.jpg', image)

    # Print the number of circles detected
    num_circles = 0 if circles is None else len(circles)
    print(f"Number of circles detected in {image_path}: {num_circles}")

即使调整Hough Circle Transform的参数，检测出的圆形仍近乎随机，恳请技术帮助！

问题分析与优化方案 #

毛孔属于小尺寸、低对比度的圆形结构，直接用霍夫圆变换很难精准检测——原图的皮肤纹理、噪声会干扰边缘检测，且你当前的参数完全不符合毛孔的实际尺寸。以下是针对性的优化步骤：

1. 预处理优化：增强毛孔对比度 #

霍夫圆对边缘清晰度要求极高，先通过自适应直方图均衡化增强局部细节，再用中值滤波保留边缘同时去噪：

# 替换原灰度+模糊步骤
img_gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# 自适应直方图均衡化，提升局部对比度
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))
img_clahe = clahe.apply(img_gray)
# 中值滤波比高斯滤波更适合保留边缘
img_blur = cv2.medianBlur(img_clahe, 3)

2. 霍夫圆参数针对性调整 #

毛孔的实际半径通常在2-8像素左右，你之前设置的minRadius=10直接过滤了所有毛孔，同时调整其他参数适配小尺寸圆形：

circles = cv2.HoughCircles(
    img_blur, 
    cv2.HOUGH_GRADIENT, 
    dp=1.2,  # 略大于1的累加器分辨率，提升检测精度
    minDist=10,  # 毛孔间距小，缩小最小距离阈值
    param1=30,  # 降低Canny边缘检测的高阈值，捕捉毛孔弱边缘
    param2=15,  # 降低累加器阈值，识别更多候选圆
    minRadius=2,  # 匹配毛孔实际最小半径
    maxRadius=8   # 匹配毛孔实际最大半径
)

3. 后过滤：筛选真实毛孔 #

检测后通过圆形度过滤误检（毛孔接近正圆，圆形度接近1）：

if circles is not None:
    circles = np.round(circles[0, :]).astype(int)
    valid_circles = []
    for (x, y, r) in circles:
        # 创建圆掩码，计算轮廓圆形度
        mask = np.zeros_like(img_gray)
        cv2.circle(mask, (x,y), r, 255, -1)
        contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
        area = cv2.contourArea(contours[0])
        perimeter = cv2.arcLength(contours[0], True)
        circularity = 4 * np.pi * (area / (perimeter ** 2)) if perimeter != 0 else 0
        # 只保留圆形度大于0.7的区域
        if circularity > 0.7:
            valid_circles.append((x,y,r))
    # 绘制有效检测结果
    for (x,y,r) in valid_circles:
        cv2.circle(image, (x,y), r, (0,0,255), 2)
    num_circles = len(valid_circles)

替代方案：基于轮廓的检测 #

如果霍夫圆效果仍不理想，可以尝试先检测轮廓再筛选圆形结构，更灵活适配毛孔特征：

# 预处理后做边缘检测
edges = cv2.Canny(img_blur, 20, 60)
# 提取外部轮廓
contours, _ = cv2.findContours(edges, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
for cnt in contours:
    area = cv2.contourArea(cnt)
    # 过滤面积不符合毛孔的区域
    if 10 < area < 50:
        perimeter = cv2.arcLength(cnt, True)
        circularity = 4 * np.pi * (area / (perimeter ** 2)) if perimeter != 0 else 0
        if circularity > 0.7:
            # 拟合最小外接圆
            (x,y), r = cv2.minEnclosingCircle(cnt)
            cv2.circle(image, (int(x),int(y)), int(r), (0,255,0), 2)

内容的提问来源于stack exchange，提问作者Eduardo