[已解决]将其中的canny检测算子，更改为sobel检测算子。请在原代码更改谢谢

Justheartyoung

import cv2
import numpy as np
# 1. 分割遥感影像为多块图幅
def split_image(image, block_size):
    height, width = image.shape[:2]
    block_images = []
    for i in range(0, height, block_size):
        for j in range(0, width, block_size):
            block = image[i:i+block_size, j:j+block_size]
            block_images.append(block)
    return block_images

# 2. 分别对每块图幅转为灰度图像
def convert_to_gray(image):
    gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    return gray_image

# 3. 对各个图像进行边缘检测
def edge_detection(image):
    edges = cv2.Canny(image, 0.5, 2)
    return edges


# 4. 消除噪声干扰
def remove_noise(image):
    denoised_image = cv2.medianBlur(image, 5)
    return denoised_image

# 5. 将各个图幅合并为一个图幅

def merge_images(images, block_size):
    num_blocks = len(images)
    rows = int(np.ceil(np.sqrt(num_blocks)))  # 向上取整
    cols = rows
    merged_image = np.zeros((rows * block_size, cols * block_size), dtype=np.uint8)
    for i, image in enumerate(images):
        x = (i % cols) * block_size
        y = (i // cols) * block_size
        merged_image[y:y+image.shape[0], x:x+image.shape[1]] = image
    return merged_image

# 6. 导出结果
def export_result(image, file_path):
    cv2.imwrite(file_path, image)

# 示例代码
image_path = r"C:\Users\WINDOWS\Desktop\taiyuan\python\dune\dune.tif"
block_size = 800

# 读取遥感影像
image = cv2.imread(image_path)

# 1. 分割遥感影像为多块图幅
block_images = split_image(image, block_size)

# 2. 分别对每块图幅转为灰度图像
gray_images = [convert_to_gray(block) for block in block_images]

# 3. 对各个图像进行边缘检测
edges_images = [edge_detection(gray) for gray in gray_images]

# 4. 消除噪声干扰
denoised_images = [remove_noise(edges) for edges in edges_images]

# 5. 将各个图幅合并为一个图幅
merged_image = merge_images(denoised_images, block_size)

# 6. 导出结果
export_result(merged_image, "output.jpg")

最佳答案

月排行榜 / 总排行榜

学习编程中的Ben

2023-8-7 16:57:57

1. import cv2
2. import numpy as np
4. # 1. 分割遥感影像为多块图幅
5. def split_image(image, block_size):
6.     height, width = image.shape[:2]
7.     block_images = []
8.     for i in range(0, height, block_size):
9.         for j in range(0, width, block_size):
10.             block = image[i:i+block_size, j:j+block_size]
11.             block_images.append(block)
12.     return block_images
14. # 2. 分别对每块图幅转为灰度图像
15. def convert_to_gray(image):
16.     gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
17.     return gray_image
19. # 3. 对各个图像进行边缘检测（使用Sobel算子）
20. def edge_detection(image):
21.     sobel_x = cv2.Sobel(image, cv2.CV_64F, 1, 0, ksize=3)
22.     sobel_y = cv2.Sobel(image, cv2.CV_64F, 0, 1, ksize=3)
23.     edges = cv2.magnitude(sobel_x, sobel_y)
24.     edges = cv2.convertScaleAbs(edges)
25.     return edges
27. # 4. 消除噪声干扰
28. def remove_noise(image):
29.     denoised_image = cv2.medianBlur(image, 5)
30.     return denoised_image
32. # 5. 将各个图幅合并为一个图幅
33. def merge_images(images, block_size):
34.     num_blocks = len(images)
35.     rows = int(np.ceil(np.sqrt(num_blocks)))  # 向上取整
36.     cols = rows
37.     merged_image = np.zeros((rows * block_size, cols * block_size), dtype=np.uint8)
38.     for i, image in enumerate(images):
39.         x = (i % cols) * block_size
40.         y = (i // cols) * block_size
41.         merged_image[y:y+image.shape[0], x:x+image.shape[1]] = image
42.     return merged_image
44. # 6. 导出结果
45. def export_result(image, file_path):
46.     cv2.imwrite(file_path, image)
48. # 示例代码
49. image_path = r"C:\Users\WINDOWS\Desktop\taiyuan\python\dune\dune.tif"
50. block_size = 800
52. # 读取遥感影像
53. image = cv2.imread(image_path)
55. # 1. 分割遥感影像为多块图幅
56. block_images = split_image(image, block_size)
58. # 2. 分别对每块图幅转为灰度图像
59. gray_images = [convert_to_gray(block) for block in block_images]
61. # 3. 对各个图像进行边缘检测
62. edges_images = [edge_detection(gray) for gray in gray_images]
64. # 4. 消除噪声干扰
65. denoised_images = [remove_noise(edges) for edges in edges_images]
67. # 5. 将各个图幅合并为一个图幅
68. merged_image = merge_images(denoised_images, block_size)
70. # 6. 导出结果
71. export_result(merged_image, "output.jpg")

复制代码

注意，在第3步中，我将边缘检测算子从Canny改为了Sobel。使用Sobel算子可以获得基于梯度的边缘信息。

如果回答对你有帮助，请给我一个最佳答案！

跳转到最佳答案楼层

学习编程中的Ben

1. import cv2
2. import numpy as np
4. # 1. 分割遥感影像为多块图幅
5. def split_image(image, block_size):
6.     height, width = image.shape[:2]
7.     block_images = []
8.     for i in range(0, height, block_size):
9.         for j in range(0, width, block_size):
10.             block = image[i:i+block_size, j:j+block_size]
11.             block_images.append(block)
12.     return block_images
14. # 2. 分别对每块图幅转为灰度图像
15. def convert_to_gray(image):
16.     gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
17.     return gray_image
19. # 3. 对各个图像进行边缘检测（使用Sobel算子）
20. def edge_detection(image):
21.     sobel_x = cv2.Sobel(image, cv2.CV_64F, 1, 0, ksize=3)
22.     sobel_y = cv2.Sobel(image, cv2.CV_64F, 0, 1, ksize=3)
23.     edges = cv2.magnitude(sobel_x, sobel_y)
24.     edges = cv2.convertScaleAbs(edges)
25.     return edges
27. # 4. 消除噪声干扰
28. def remove_noise(image):
29.     denoised_image = cv2.medianBlur(image, 5)
30.     return denoised_image
32. # 5. 将各个图幅合并为一个图幅
33. def merge_images(images, block_size):
34.     num_blocks = len(images)
35.     rows = int(np.ceil(np.sqrt(num_blocks)))  # 向上取整
36.     cols = rows
37.     merged_image = np.zeros((rows * block_size, cols * block_size), dtype=np.uint8)
38.     for i, image in enumerate(images):
39.         x = (i % cols) * block_size
40.         y = (i // cols) * block_size
41.         merged_image[y:y+image.shape[0], x:x+image.shape[1]] = image
42.     return merged_image
44. # 6. 导出结果
45. def export_result(image, file_path):
46.     cv2.imwrite(file_path, image)
48. # 示例代码
49. image_path = r"C:\Users\WINDOWS\Desktop\taiyuan\python\dune\dune.tif"
50. block_size = 800
52. # 读取遥感影像
53. image = cv2.imread(image_path)
55. # 1. 分割遥感影像为多块图幅
56. block_images = split_image(image, block_size)
58. # 2. 分别对每块图幅转为灰度图像
59. gray_images = [convert_to_gray(block) for block in block_images]
61. # 3. 对各个图像进行边缘检测
62. edges_images = [edge_detection(gray) for gray in gray_images]
64. # 4. 消除噪声干扰
65. denoised_images = [remove_noise(edges) for edges in edges_images]
67. # 5. 将各个图幅合并为一个图幅
68. merged_image = merge_images(denoised_images, block_size)
70. # 6. 导出结果
71. export_result(merged_image, "output.jpg")

复制代码

注意，在第3步中，我将边缘检测算子从Canny改为了Sobel。使用Sobel算子可以获得基于梯度的边缘信息。

如果回答对你有帮助，请给我一个最佳答案！

陶远航

在原代码中，将Canny边缘检测算子更改为Sobel边缘检测算子，请看下面的修改后的代码：

1. import cv2
   
2. import numpy as np
   
3. 
   
4. # 1. 分割遥感影像为多块图幅
   
5. def split_image(image, block_size):
   
6.     height, width = image.shape[:2]
   
7.     block_images = []
   
8.     for i in range(0, height, block_size):
   
9.         for j in range(0, width, block_size):
   
10.             block = image[i:i+block_size, j:j+block_size]
    
11.             block_images.append(block)
    
12.     return block_images
    
13. 
    
14. # 2. 分别对每块图幅转为灰度图像
    
15. def convert_to_gray(image):
    
16.     gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    
17.     return gray_image
    
18. 
    
19. # 3. 对各个图像进行边缘检测
    
20. def edge_detection(image):
    
21.     sobel_x = cv2.Sobel(image, cv2.CV_64F, 1, 0, ksize=3)
    
22.     sobel_y = cv2.Sobel(image, cv2.CV_64F, 0, 1, ksize=3)
    
23.     edges = np.sqrt(np.square(sobel_x) + np.square(sobel_y))
    
24.     edges = edges.astype(np.uint8)
    
25.     return edges
    
26. 
    
27. # 4. 消除噪声干扰
    
28. def remove_noise(image):
    
29.     denoised_image = cv2.medianBlur(image, 5)
    
30.     return denoised_image
    
31. 
    
32. # 5. 将各个图幅合并为一个图幅
    
33. def merge_images(images, block_size):
    
34.     num_blocks = len(images)
    
35.     rows = int(np.ceil(np.sqrt(num_blocks)))  # 向上取整
    
36.     cols = rows
    
37.     merged_image = np.zeros((rows * block_size, cols * block_size), dtype=np.uint8)
    
38.     for i, image in enumerate(images):
    
39.         x = (i % cols) * block_size
    
40.         y = (i // cols) * block_size
    
41.         merged_image[y:y+image.shape[0], x:x+image.shape[1]] = image
    
42.     return merged_image
    
43. 
    
44. # 6. 导出结果
    
45. def export_result(image, file_path):
    
46.     cv2.imwrite(file_path, image)
    
47. 
    
48. # 示例代码
    
49. image_path = r"C:\Users\WINDOWS\Desktop\taiyuan\python\dune\dune.tif"
    
50. block_size = 800
    
51. 
    
52. # 读取遥感影像
    
53. image = cv2.imread(image_path)
    
54. 
    
55. # 1. 分割遥感影像为多块图幅
    
56. block_images = split_image(image, block_size)
    
57. 
    
58. # 2. 分别对每块图幅转为灰度图像
    
59. gray_images = [convert_to_gray(block) for block in block_images]
    
60. 
    
61. # 3. 对各个图像进行边缘检测
    
62. edges_images = [edge_detection(gray) for gray in gray_images]
    
63. 
    
64. # 4. 消除噪声干扰
    
65. denoised_images = [remove_noise(edges) for edges in edges_images]
    
66. 
    
67. # 5. 将各个图幅合并为一个图幅
    
68. merged_image = merge_images(denoised_images, block_size)
    
69. 
    
70. # 6. 导出结果
    
71. export_result(merged_image, "output.jpg")
    

复制代码

这段修改后的代码将使用Sobel边缘检测算子进行图像边缘检测。希望对你有所帮助！如果还有其他问题，请随时提问。
如果问题已经解决，请设置最佳答案

isdkz

import cv2
import numpy as np

# 1. 分割遥感影像为多块图幅
def split_image(image, block_size):
    height, width = image.shape[:2]
    block_images = []
    for i in range(0, height, block_size):
        for j in range(0, width, block_size):
            block = image[i:i+block_size, j:j+block_size]
            block_images.append(block)
    return block_images

# 2. 分别对每块图幅转为灰度图像
def convert_to_gray(image):
    gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    return gray_image

# 3. 对各个图像进行边缘检测
def edge_detection(image):
    edges = cv2.Sobel(image, cv2.CV_64F, 1, 1, ksize=3)
    edges = cv2.convertScaleAbs(edges)
    return edges

# 4. 消除噪声干扰
def remove_noise(image):
    denoised_image = cv2.medianBlur(image, 5)
    return denoised_image

# 5. 将各个图幅合并为一个图幅
def merge_images(images, block_size):
    num_blocks = len(images)
    rows = int(np.ceil(np.sqrt(num_blocks)))  # 向上取整
    cols = rows
    merged_image = np.zeros((rows * block_size, cols * block_size), dtype=np.uint8)
    for i, image in enumerate(images):
        x = (i % cols) * block_size
        y = (i // cols) * block_size
        merged_image[y:y+image.shape[0], x:x+image.shape[1]] = image
    return merged_image

# 6. 导出结果
def export_result(image, file_path):
    cv2.imwrite(file_path, image)

# 示例代码
image_path = r"C:\Users\WINDOWS\Desktop\taiyuan\python\dune\dune.tif"
block_size = 800

# 读取遥感影像
image = cv2.imread(image_path)

# 1. 分割遥感影像为多块图幅
block_images = split_image(image, block_size)

# 2. 分别对每块图幅转为灰度图像
gray_images = [convert_to_gray(block) for block in block_images]

# 3. 对各个图像进行边缘检测
edges_images = [edge_detection(gray) for gray in gray_images]

# 4. 消除噪声干扰
denoised_images = [remove_noise(edges) for edges in edges_images]

# 5. 将各个图幅合并为一个图幅
merged_image = merge_images(denoised_images, block_size)

# 6. 导出结果
export_result(merged_image, "output.jpg")

将边缘检测算子从Canny更改为Sobel的方法是使用cv2.Sobel函数来进行边缘检测。在原代码中的edge_detection函数中，将cv2.Canny替换为cv2.Sobel，并指定相关参数即可。Sobel函数的第一个参数是要进行边缘检测的图像，第二个参数是输出图像的数据类型，第三和第四个参数是指定x和y方向的导数的阶数，第五个参数是指定卷积核的大小。在本例中，我们使用了3x3的卷积核。最后，我们使用cv2.convertScaleAbs函数将结果转换为无符号8位整数类型。