谁能把图片中的很淡的脏污抓出来

附件图片中的脏污试了很多办法，没有抓出来，求高人来试一下，刚注册，没money，见谅！

1. read_image (Image, 'C:/Users/jackgu/Desktop/脏污图片/1.bmp')
   
2. get_image_size (Image, Width, Height)
   
3. Sigma1 := 30.0
   
4. Sigma2 := 5.0
   
5. gen_gauss_filter (GaussFilter1, Sigma1, Sigma1, 0.0, 'none', 'rft', Width, Height)
   
6. gen_gauss_filter (GaussFilter2, Sigma2, Sigma2, 0.0, 'none', 'rft', Width, Height)
   
7. sub_image (GaussFilter1, GaussFilter2, Filter, 1, 0)
   
8. 
   
9. rgb1_to_gray (Image, GrayImage)
   
10. rft_generic (GrayImage, ImageFFT, 'to_freq', 'none', 'complex', Width)
    
11. convol_fft (ImageFFT, Filter, ImageConvol)
    
12. rft_generic (ImageConvol, ImageFiltered, 'from_freq', 'n', 'real', Width)
    
13. scale_image_range (ImageFiltered, ImageScaled, 0, 255)
    
14. threshold (ImageScaled, Region, 1, 255)
    
15. connection (Region, ConnectedRegions)
    
16. select_shape_proto (ConnectedRegions, ConnectedRegions, SelectedRegions, 'distance_contour', 1, 999)
    
17. opening_circle (SelectedRegions, RegionOpening, 7.5)
    
18. 
    
19. dev_display (Image)
    
20. dev_display (RegionOpening)

复制代码

1. dev_set_draw ('margin')
   
2. read_image (Image, '3.jpg')
   
3. rgb1_to_gray (Image, GrayImage)
   
4. emphasize (GrayImage, ImageEmphasize3, 500, 500, 50)
   
5. threshold (ImageEmphasize3, Region1, 0, 150)
   
6. closing_circle (Region1, RegionClosing2, 5)
   
7. opening_circle (RegionClosing2, RegionOpening1, 5)
   
8. connection (RegionOpening1, ConnectedRegions3)
   
9. select_shape (ConnectedRegions3, SelectedRegions3, 'area', 'and', 100000, 200000)
   
10. closing_circle (SelectedRegions3, RegionClosing3, 10)
    
11. fill_up (RegionClosing3, RegionFillUp)
    
12. dev_display (Image)
    
13. shape_trans (RegionFillUp, RegionTrans, 'outer_circle')

复制代码

大神，按照你的方法确实能抓出来，我又上传了2张，有没有办法兼顾一下？

:handshake

学习了！

1. dev_set_draw ('margin')
   
2. read_image (Image, '1.bmp')
   
3. rgb1_to_gray (Image, GrayImage)
   
4. emphasize (GrayImage, ImageEmphasize3, 90, 90, 40)
   
5. * emphasize (GrayImage, ImageEmphasize3, 500,500, 100)
   
6. threshold (ImageEmphasize3, Regions, 0, 179)
   
7. connection (Regions, ConnectedRegions)
   
8. * threshold (ImageEmphasize3, Region1, 0, 150)
   
9. closing_circle (ConnectedRegions, RegionClosing2, 10)
   
10. opening_circle (RegionClosing2, RegionOpening1, 10)
    
11. connection (RegionOpening1, ConnectedRegions3)
    
12. select_shape (ConnectedRegions3, SelectedRegions, 'area', 'and', 3000, 41293.2)
    
13. * select_shape (ConnectedRegions3, SelectedRegions3, 'area', 'and', 100000, 200000)
    
14. closing_circle (SelectedRegions, RegionClosing3, 10)
    
15. fill_up (RegionClosing3, RegionFillUp)
    
16. dev_display (Image)
    
17. shape_trans (RegionFillUp, RegionTrans, 'outer_circle')

复制代码

这个应该可以识别你的后面两幅图的脏污检测

学习了！