使用Claude Code高效实现图像边缘检测的实践指南

# 创建虚拟环境（可选）
python -m venv .venv

# 激活虚拟环境
.venv\Scripts\activate

# 安装依赖
pip install opencv-python matplotlib numpy

灰度值：  50  50  50  200 200 200  50  50
                   ↑        ↑
                 左边缘    右边缘
             (值从50突变到200)

pip install opencv-python matplotlib

"""
边缘检测示例
读取图片，使用多种算法进行边缘检测，对比展示结果
"""
import cv2
import numpy as np
import matplotlib.pyplot as plt
def sobel_edge(gray):
    """Sobel 边缘检测"""
    x = cv2.Sobel(gray, cv2.CV_64F, 1, 0, ksize=3)
    y = cv2.Sobel(gray, cv2.CV_64F, 0, 1, ksize=3)
    edge = np.sqrt(x ** 2 + y ** 2)
    return np.clip(edge, 0, 255).astype(np.uint8)
def canny_edge(gray, low=50, high=150):
    """Canny 边缘检测"""
    return cv2.Canny(gray, low, high)
def main():
    img_path = r'D:\Temp\testCheck.jpg'
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    blurred = cv2.GaussianBlur(gray, (5, 5), sigmaX=1.5)
    results = {
        '原图': cv2.cvtColor(img, cv2.COLOR_BGR2RGB),
        '灰度图': gray,
        'Sobel': sobel_edge(blurred),
        'Laplacian': cv2.Laplacian(blurred, cv2.CV_64F),
        'Canny (50,150)': canny_edge(blurred, 50, 150),
        'Canny (100,200)': canny_edge(blurred, 100, 200),
    }
    # 可视化对比展示...

# 修复前（中文显示异常）
matplotlib.rcParams['font.sans-serif'] = ['DejaVu Sans']
# 修复后（指定中文字体）
matplotlib.rcParams['font.sans-serif'] = ['SimHei', 'Microsoft YaHei', 'KaiTi']

main()
  │
  ├─ 读取图片 → 灰度化 → 高斯降噪
  │
  ├─ 多算法边缘检测
  │   ├─ sobel_edge()
  │   ├─ laplacian_edge()
  │   └─ canny_edge()
  │
  └─ matplotlib 2×3 可视化对比

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

blurred = cv2.GaussianBlur(gray, (5, 5), sigmaX=1.5)

def sobel_edge(gray):
    x = cv2.Sobel(gray, cv2.CV_64F, 1, 0, ksize=3)  # 水平方向梯度
    y = cv2.Sobel(gray, cv2.CV_64F, 0, 1, ksize=3)  # 垂直方向梯度
    edge = np.sqrt(x ** 2 + y ** 2)                  # 合成梯度幅值
    return np.clip(edge, 0, 255).astype(np.uint8)

水平边缘检测 Sobel X 核:     垂直边缘检测 Sobel Y 核:
-1  0  1                      -1 -2 -1
-2  0  2                       0  0  0
-1  0  1                       1  2  1

def canny_edge(gray, low=50, high=150):
    return cv2.Canny(gray, low, high)

fig, axes = plt.subplots(2, 3, figsize=(15, 10))
fig.suptitle('边缘检测算法对比', fontsize=22, fontweight='bold')

for ax, (title, image) in zip(axes.flat, results.items()):
    if len(image.shape) == 3:
        ax.imshow(image)           # 彩色图用 RGB
    else:
        ax.imshow(image, cmap='gray')  # 灰度图用灰度色阶
    ax.set_title(title, fontsize=14, fontweight='bold')
    ax.axis('off')

plt.tight_layout()
plt.savefig('edge_detection_result.png', dpi=150, bbox_inches='tight')
plt.show()

# 噪声多的图：提高低阈值，减少误检
edges = cv2.Canny(blurred, low=80, high=150)
# 细节多的图：降低低阈值，不漏检
edges = cv2.Canny(blurred, low=30, high=100)
# 高阈值差（high - low 大）→ 更少但更可靠的边缘
# 低阈值差（high - low 小）→ 更多但可能包含噪声的边缘

# 找到轮廓后，计算像素距离，再根据标定系数转换为物理尺寸
contours, _ = cv2.findContours(edges, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnt = contours[0]
x, y, w, h = cv2.boundingRect(cnt)
real_width = w * px_to_mm  # 像素 → 毫米

import pyzbar
barcodes = pyzbar.decode(img)
for barcode in barcodes:
    product_id = barcode.data.decode('utf-8')
    tolerance = load_tolerance(product_id)

"""
边缘检测示例
读取图片，使用多种算法进行边缘检测，对比展示结果
"""

import cv2
import numpy as np
import matplotlib
import matplotlib.pyplot as plt

# 设置中文字体
matplotlib.rcParams['font.sans-serif'] = ['SimHei', 'Microsoft YaHei', 'KaiTi']
matplotlib.rcParams['axes.unicode_minus'] = False


def sobel_edge(gray):
    """Sobel 边缘检测"""
    x = cv2.Sobel(gray, cv2.CV_64F, 1, 0, ksize=3)
    y = cv2.Sobel(gray, cv2.CV_64F, 0, 1, ksize=3)
    edge = np.sqrt(x ** 2 + y ** 2)
    return np.clip(edge, 0, 255).astype(np.uint8)


def laplacian_edge(gray):
    """Laplacian 边缘检测"""
    return cv2.Laplacian(gray, cv2.CV_64F, ksize=3)


def canny_edge(gray, low=50, high=150):
    """Canny 边缘检测"""
    return cv2.Canny(gray, low, high)


def main():
    # ========== 1. 读取图片 ==========
    img_path = r'D:\Temp\testCheck.jpg'
    output_path = r'D:\Temp\edge_detection_result.png'

    img = cv2.imread(img_path)
    if img is None:
        print(f"错误：无法读取图片 {img_path}")
        return

    h, w = img.shape[:2]
    print(f"图片尺寸: {w} x {h}")

    # 转灰度图（彩色转灰度：边缘检测只关心亮度变化，转灰度能减少 2/3 的计算量）
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

    # 高斯模糊降噪（提前过滤掉噪点，防止它们被误判为边缘）
    blurred = cv2.GaussianBlur(gray, (5, 5), sigmaX=1.5)

    # ========== 2. 多种边缘检测 ==========
    results = {
        '原图': cv2.cvtColor(img, cv2.COLOR_BGR2RGB),
        '灰度图': gray,
        'Sobel': sobel_edge(blurred),
        'Laplacian': laplacian_edge(blurred),
        'Canny (50,150)': canny_edge(blurred, 50, 150),
        'Canny (100,200)': canny_edge(blurred, 100, 200),
    }

    # ========== 3. 可视化对比 ==========
    fig, axes = plt.subplots(2, 3, figsize=(15, 10))
    fig.suptitle('边缘检测算法对比', fontsize=22, fontweight='bold', color='#333333')

    titles = list(results.keys())
    images = list(results.values())

    for idx, (ax, title, image) in enumerate(zip(axes.flat, titles, images)):
        # 彩色图用 RGB 显示，灰度图用灰度显示
        if len(image.shape) == 3:
            ax.imshow(image)
        else:
            ax.imshow(image, cmap='gray')

        ax.set_title(title, fontsize=14, fontweight='bold')
        ax.axis('off')

    plt.tight_layout(rect=[0, 0, 1, 0.95]) # 自动调整子图间距，防止标题重叠
    plt.savefig(output_path, dpi=150, bbox_inches='tight')# 将对比结果高清保存到本地
    plt.show()# 弹窗展示

    print(f"结果已保存到 {output_path}")


if __name__ == '__main__':
    main()