完成高斯高通滤波

Filtering_Gaussian_Highpass.py

@@ -0,0 +1,63 @@
+import cv2
+import numpy as np
+import matplotlib.pyplot as plt
+plt.rcParams['font.sans-serif'] = ['simHei']
+plt.rcParams['axes.unicode_minus'] = False
+from Gaussian_Highpass import gaussian_high_pass_filter
+
+# 读取照片
+image_characterTestPattern2 = cv2.imread('file/characterTestPattern2.jpg', 0)  # 改成你自己的图片路径
+
+# 快速傅里叶变换
+dft_characterTestPattern2 = np.fft.fft2(image_characterTestPattern2)
+shift_characterTestPattern2 = np.fft.fftshift(dft_characterTestPattern2)
+
+# 高斯高通滤波器参数
+D0 = 10
+D1 = 100
+D2 = 1000
+ghpf0 = gaussian_high_pass_filter(image_characterTestPattern2.shape, D0)
+ghpf1 = gaussian_high_pass_filter(image_characterTestPattern2.shape, D1)
+ghpf2 = gaussian_high_pass_filter(image_characterTestPattern2.shape, D2)
+
+# 应用滤波器
+shift_filtered0 = shift_characterTestPattern2 * ghpf0
+shift_filtered1 = shift_characterTestPattern2 * ghpf1
+shift_filtered2 = shift_characterTestPattern2 * ghpf2
+
+# 逆傅里叶变换
+dft_ishift0 = np.fft.ifftshift(shift_filtered0)
+image0 = np.fft.ifft2(dft_ishift0)
+image0 = np.abs(image0)
+
+dft_ishift1 = np.fft.ifftshift(shift_filtered1)
+image1 = np.fft.ifft2(dft_ishift1)
+image1 = np.abs(image1)
+
+dft_ishift2 = np.fft.ifftshift(shift_filtered2)
+image2 = np.fft.ifft2(dft_ishift2)
+image2 = np.abs(image2)
+
+#展示图像
+plt.subplot(2,2,1)
+plt.imshow(image_characterTestPattern2, cmap='gray')
+plt.title('原图')
+plt.axis('off')
+
+plt.subplot(2,2,2)
+plt.imshow(image0, cmap='gray')
+plt.title('高斯高通滤波器D0=10')
+plt.axis('off')
+
+plt.subplot(2,2,3)
+plt.imshow(image1, cmap='gray')
+plt.title('高斯高通滤波器D0=100')
+plt.axis('off')
+
+plt.subplot(2,2,4)
+plt.imshow(image2, cmap='gray')
+plt.title('高斯高通滤波器D0=1000')
+plt.axis('off')
+
+plt.tight_layout()
+plt.show()

Gaussian_Highpass.py

@@ -0,0 +1,11 @@
+import numpy as np
+
+def gaussian_high_pass_filter(shape, cutoff):
+    rows, cols = shape
+    crow, ccol = rows // 2 , cols // 2
+    x = np.arange(0, cols)
+    y = np.arange(0, rows)
+    x, y = np.meshgrid(x, y)
+    distance = np.sqrt((x - ccol)**2 + (y - crow)**2)
+    high_pass = 1 - np.exp(-(distance**2) / (2 * (cutoff**2)))
+    return high_pass