Saving Images after calculating local entropy

Image Analysis
#1

Hi,
I am trying to save the images after calculation local entropy.
I followed https://scikit-image.org/docs/stable/auto_examples/filters/plot_entropy.html.
I used cv2.imwrite. I saved black images. I tried with io.imsave. It is saving the image in low contrast.
Below is my code.

for root, dirs, file in os.walk(image_path, topdown = False):
for name in dirs:
dir = root + os.sep + name
path = os.path.basename(dir)
for filename in os.listdir(dir):
file = dir + os.sep + filename
image = io.imread(file, plugin = ‘matplotlib’)

        im = (rgb2gray(image))
        print(type(im))
        ent_image = entropy(im, disk(10))
        img = resize(ent_image, (299,299))
        
        image_saved_path = saved_path + os.sep + path
        if not os.path.exists(image_saved_path ):
           os.makedirs(image_saved_path)
                     
        image_base = os.path.basename(file)
     
      
       # io.imsave(image_saved_path + os.sep + image_base, img.astype(np.uint8))
       # img /= 255.0
        cv2.imwrite(image_saved_path + os.sep + image_base,img_as_ubyte(img))

Please help me out to solve the issue.

1 Like
#2

Hi, could you check the values in the img image before saving?
Maybe they are outside of the unsigned byte datatype in which you are trying to save the image?

#3

Yes, as @Christian_Tischer says, it’s likely that the entropy image doesn’t have the expected range for a floating point image in scikit-image, or has a restricted range. For more on image intensity ranges in scikit-image, see this document:

https://scikit-image.org/docs/dev/user_guide/data_types.html

The solution is to rescale your image to whatever output range you want. The function skimage.exposure.rescale_intensity can make this convenient:

https://scikit-image.org/docs/dev/api/skimage.exposure.html#skimage.exposure.rescale_intensity

1 Like