环境准备
pip install opencv-python pillow numpy matplotlib scikit-image
图像读取与显示
import cv2
import numpy as np
from PIL import Image
import matplotlib.pyplot as plt
# 读取图像
img = cv2.imread('image.jpg')
# OpenCV使用BGR格式，转换为RGB用于显示
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# 显示图像
plt.imshow(img_rgb)
plt.axis('off')
plt.show()
几何变换示例
def resize_image(img_path, new_width, new_height):
img = cv2.imread(img_path)
resized = cv2.resize(img, (new_width, new_height), interpolation=cv2.INTER_LINEAR)
return resized
def rotate_image(img, angle):
height, width = img.shape[:2]
center = (width//2, height//2)
rotation_matrix = cv2.getRotationMatrix2D(center, angle, 1.0)
rotated = cv2.warpAffine(img, rotation_matrix, (width, height))
return rotated
滤波算法示例
def apply_filters(img_path):
img = cv2.imread(img_path)
# 均值滤波
mean_filtered = cv2.blur(img, (5, 5))
# 高斯滤波
gaussian_filtered = cv2.GaussianBlur(img, (5, 5), 0)
# 中值滤波
median_filtered = cv2.medianBlur(img, 5)
# 双边滤波（保边去噪）
bilateral_filtered = cv2.bilateralFilter(img, 9, 75, 75)
return {
'original': img,
'mean': mean_filtered,
'gaussian': gaussian_filtered,
'median': median_filtered,
'bilateral': bilateral_filtered
}
边缘检测示例
def edge_detection(img_path):
img = cv2.imread(img_path, 0) # 读取为灰度图
# Sobel 边缘检测
sobel_x = cv2.Sobel(img, cv2.CV_64F, 1, 0, ksize=3)
sobel_y = cv2.Sobel(img, cv2.CV_64F, 0, 1, ksize=3)
sobel_combined = np.sqrt(sobel_x**2 + sobel_y**2)
# Canny 边缘检测
canny_edges = cv2.Canny(img, threshold1=50, threshold2=150)
# Laplacian 边缘检测
laplacian = cv2.Laplacian(img, cv2.CV_64F)
return {
'sobel': sobel_combined,
'canny': canny_edges,
'laplacian': laplacian
}
形态学操作示例
def morphological_operations(img_path):
img = cv2.imread(img_path, 0) # 读取为灰度图
# 定义结构元素
kernel = np.ones((5, 5), np.uint8)
# 腐蚀
erosion = cv2.erode(img, kernel, iterations=1)
# 膨胀
dilation = cv2.dilate(img, kernel, iterations=1)
# 开运算（先腐蚀后膨胀）
opening = cv2.morphologyEx(img, cv2.MORPH_OPEN, kernel)
# 闭运算（先膨胀后腐蚀）
closing = cv2.morphologyEx(img, cv2.MORPH_CLOSE, kernel)
# 梯度形态学
gradient = cv2.morphologyEx(img, cv2.MORPH_GRADIENT, kernel)
return {
'erosion': erosion,
'dilation': dilation,
'opening': opening,
'closing': closing,
'gradient': gradient
}
色彩空间转换示例
def color_space_conversions(img_path):
img = cv2.imread(img_path)
# BGR 转 RGB
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# BGR 转 HSV
img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
# BGR 转灰度
img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# BGR 转 LAB
img_lab = cv2.cvtColor(img, cv2.COLOR_BGR2LAB)
# BGR 转 YCrCb
img_ycrcb = cv2.cvtColor(img, cv2.COLOR_BGR2YCrCb)
return {
'rgb': img_rgb,
'hsv': img_hsv,
'gray': img_gray,
'lab': img_lab,
'ycrcb': img_ycrcb
}
直方图处理示例
def histogram_processing(img_path):
img = cv2.imread(img_path, 0) # 读取为灰度图
# 计算直方图
hist, bins = np.histogram(img.flatten(), 256, [0, 256])
# 直方图均衡化
equalized = cv2.equalizeHist(img)
# 创建 CLAHE 对象（限制对比度自适应直方图均衡化）
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
clahe_result = clahe.apply(img)
return {
'histogram': hist,
'equalized': equalized,
'clahe': clahe_result
}
阈值处理示例
def thresholding(img_path):
img = cv2.imread(img_path, 0) # 读取为灰度图
# 简单阈值
ret, thresh_binary = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)
# 自适应阈值（高斯）
adaptive_gaussian = cv2.adaptiveThreshold(
img, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY, 11, 2
)
# 自适应阈值（均值）
adaptive_mean = cv2.adaptiveThreshold(
img, 255, cv2.ADAPTIVE_THRESH_MEAN_C,
cv2.THRESH_BINARY, 11, 2
)
# Otsu 阈值
ret, otsu = cv2.threshold(img, 0, 255,
cv2.THRESH_BINARY + cv2.THRESH_OTSU)
return {
'binary': thresh_binary,
'adaptive_gaussian': adaptive_gaussian,
'adaptive_mean': adaptive_mean,
'otsu': otsu
}
轮廓检测示例
def contour_detection(img_path):
img = cv2.imread(img_path)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# 阈值处理
ret, thresh = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY)
# 查找轮廓
contours, hierarchy = cv2.findContours(
thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE
)
# 绘制轮廓
contour_img = img.copy()
cv2.drawContours(contour_img, contours, -1, (0, 255, 0), 2)
# 查找轮廓的边界框
bounding_boxes = [cv2.boundingRect(c) for c in contours]
return {
'contours': contours,
'hierarchy': hierarchy,
'contour_image': contour_img,
'bounding_boxes': bounding_boxes
}
霍夫变换直线检测示例
def hough_line_detection(img_path):
img = cv2.imread(img_path)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# 边缘检测
edges = cv2.Canny(gray, 50, 150, apertureSize=3)
# 标准霍夫变换
lines = cv2.HoughLines(edges, 1, np.pi / 180, 200)
# 概率霍夫变换
lines_prob = cv2.HoughLinesP(
edges, 1, np.pi / 180, threshold=100,
minLineLength=100, maxLineGap=10
)
# 在原图上绘制检测到的直线
result = img.copy()
if lines_prob is not None:
for line in lines_prob:
x1, y1, x2, y2 = line[0]
cv2.line(result, (x1, y1), (x2, y2), (0, 255, 0), 2)
return {
'edges': edges,
'lines': lines,
'lines_probabilistic': lines_prob,
'result': result
}
频域处理示例
def frequency_domain_processing(img_path):
img = cv2.imread(img_path, 0) # 读取为灰度图
# 快速傅里叶变换
f_transform = np.fft.fft2(img)
f_shift = np.fft.fftshift(f_transform)
# 计算幅度谱
magnitude_spectrum = 20 * np.log(np.abs(f_shift) + 1)
# 创建高通滤波器
rows, cols = img.shape
crow, ccol = rows // 2, cols // 2
mask = np.ones((rows, cols), np.uint8)
r, c = 30, 30 # 截止频率
mask[crow-r:crow+r, ccol-r:ccol+r] = 0
# 应用高通滤波
f_shift_masked = f_shift * mask
f_ishift = np.fft.ifftshift(f_shift_masked)
img_back = np.fft.ifft2(f_ishift)
img_back = np.abs(img_back)
return {
'magnitude_spectrum': magnitude_spectrum,
'highpass_result': img_back,
'mask': mask
}
特征提取示例
def feature_extraction(img_path):
from skimage import feature, measure
img = cv2.imread(img_path, 0) # 读取为灰度图
# Canny 边缘检测
edges = feature.canny(img, sigma=1)
# Harris 角点检测
corners = feature.corner_harris(img)
corner_coords = feature.corner_peaks(corners, min_distance=5)
# SIFT 特征点检测
sift = cv2.SIFT_create()
keypoints, descriptors = sift.detectAndCompute(img, None)
# ORB 特征点检测
orb = cv2.ORB_create()
orb_keypoints, orb_descriptors = orb.detectAndCompute(img, None)
# 绘制特征点
feature_img = cv2.drawKeypoints(
img, keypoints, None,
flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS
)
return {
'edges': edges,
'corners': corner_coords,
'sift_keypoints': keypoints,
'sift_descriptors': descriptors,
'orb_keypoints': orb_keypoints,
'feature_image': feature_img
}
图像金字塔与多尺度检测示例
def image_pyramid(img_path):
img = cv2.imread(img_path)
# 高斯金字塔
gaussian_pyramid = [img]
for i in range(4):
img = cv2.pyrDown(img)
gaussian_pyramid.append(img)
# 拉普拉斯金字塔
laplacian_pyramid = []
for i in range(len(gaussian_pyramid) - 1, 0, -1):
expanded = cv2.pyrUp(gaussian_pyramid[i])
laplacian = cv2.subtract(gaussian_pyramid[i-1], expanded)
laplacian_pyramid.append(laplacian)
return {
'gaussian': gaussian_pyramid,
'laplacian': laplacian_pyramid
}
完整示例：综合图像处理流程
def complete_image_processing_pipeline(img_path):
"""
完整的图像处理流程示例
"""
# 1. 读取图像
img = cv2.imread(img_path)
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# 2. 预处理 - 去噪
denoised = cv2.bilateralFilter(img, 9, 75, 75)
# 3. 转换为灰度图
gray = cv2.cvtColor(denoised, cv2.COLOR_BGR2GRAY)
# 4. 增强对比度 - 直方图均衡化
equalized = cv2.equalizeHist(gray)
# 5. 边缘检测
edges = cv2.Canny(equalized, 50, 150)
# 6. 形态学操作 - 填充小孔洞
kernel = np.ones((3, 3), np.uint8)
dilated_edges = cv2.dilate(edges, kernel, iterations=2)
eroded_edges = cv2.erode(dilated_edges, kernel, iterations=1)
# 7. 查找轮廓
contours, _ = cv2.findContours(
eroded_edges, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE
)
# 8. 绘制结果
result = img_rgb.copy()
cv2.drawContours(result, contours, -1, (0, 255, 0), 2)
return {
'original': img_rgb,
'denoised': denoised,
'gray': gray,
'equalized': equalized,
'edges': edges,
'processed_edges': eroded_edges,
'result': result,
'contour_count': len(contours)
}
# 使用示例
if __name__ == "__main__":
result = complete_image_processing_pipeline('sample.jpg')
print(f"检测到 {result['contour_count']} 个轮廓")
# 显示结果
import matplotlib.pyplot as plt
plt.figure(figsize=(15, 10))
plt.subplot(2, 4, 1)
plt.imshow(result['original'])
plt.title('原始图像')
plt.subplot(2, 4, 4)
plt.imshow(result['result'])
plt.title('最终结果')
plt.tight_layout()
plt.show()