3DGabor滤波器实现人脸特性提取

1. import cv2
2. import numpy as np
4. # 定义 Gabor 滤波器的参数
5. kSize = 31  # 滤波器核的大小
6. g_sigma = 3.0  # 高斯包络的标准差
7. g_theta = np.pi / 4  # Gabor 函数的方向
8. g_lambda = 10.0  # 正弦波的波长
9. g_gamma = 0.5  # 空间纵横比
10. g_psi = np.pi / 2  # 相位偏移
12. # 生成 Gabor 滤波器核
13. kernel = cv2.getGaborKernel((kSize, kSize), g_sigma, g_theta, g_lambda, g_gamma, g_psi, ktype=cv2.CV_32F)
15. # 读取多波段图像
16. image = cv2.imread('1.png', cv2.IMREAD_UNCHANGED)
18. # 获取图像的波段数
19. num_bands = image.shape[2] if len(image.shape) == 3 else 1
21. # 初始化处理后的多波段图像
22. filtered_image = np.zeros_like(image, dtype=np.float32)
24. # 遍历每个波段
25. for band in range(num_bands):
26.     # 提取当前波段
27.     band_image = image[:, :, band] if len(image.shape) == 3 else image
29.     # 应用 Gabor 滤波器
30.     filtered_band_image = cv2.filter2D(band_image, cv2.CV_32F, kernel)
32.     # 将处理后的波段放回结果图像中
33.     if len(image.shape) == 3:
34.         filtered_image[:, :, band] = filtered_band_image
35.     else:
36.         filtered_image = filtered_band_image
38. # 将处理后的图像转换为合适的数据类型
39. filtered_image = np.clip(filtered_image, 0, 255).astype(np.uint8)
41. # 显示结果
42. if num_bands == 3:
43.     cv2.imshow('Original Image', image)
44.     cv2.imshow('Filtered Image', filtered_image)
45. else:
46.     for band in range(num_bands):
47.         cv2.imshow(f'Original Band {band + 1}', image[:, :, band] if len(image.shape) == 3 else image)
48.         cv2.imshow(f'Filtered Band {band + 1}', filtered_image[:, :, band] if len(image.shape) == 3 else filtered_image)
50. # 保存结果
51. cv2.imwrite('filtered_multiband_image.png', filtered_image)
53. # 等待按键，然后关闭所有窗口
54. cv2.waitKey(0)
55. cv2.destroyAllWindows()

复制代码