I can't thank you enough!! I'm an electrochemist venturing into microscopy and image processing. These are the best introduction to microscopic image processing with python
I am a TEM person and learned a lot from your videos. I am applying what I have learned from you to my research. Thank you for taking the time and effort to make these amazing videos :)
I want to do area analysis of particles and phase analysis from an SEM micrograph. I'm facing a problem in separating particles having overlapping areas. Though it identifies individual particles, it also takes their cluster as another separate entity. Any video on that ? I'm new to the channel.
Thanks for sharing. I am looking for filter or denoising grains image as shared in this video example but also has stress release fractures that cut each other and make close boundaries and resulting in misleading grain size…? I hope I explain my question well.
Thank you for the fantastic tutorials. I have watched tutorial 16 to 26. These tutorials deal with 2D images. My project uses 3D tif. Do the functions in these tutorials also work for 3D images? Do you have a guide about it?
I will record a video on how to work with 3D images. Until then, here is some info... 2D images have numpy array shape as (x, y) for gray images and (x, y, RGB) for color images. 3D images typically have structure (z, x, y) for gray and (z, x, y, RGB) for color image stacks. Most opencv and scikit-image functions are designed for 2D images. Therefore you need to read one slice at a time inside a for loop and apply the image processing function. Easier than it sounds... If I have an image of size (50, 256, 256) - 50 slices gray. If I want to apply threshold to the entire volume, slice by slice. binary_img = [] for img in range(image.shape[0]: #This would be 50 in our example input_img = image[img, :, :] #Define temporary image slice name thresh = 75 #Or use otsu binary = input_img > thresh binary_img.append(binary) #Add binarized slice to our placeholder list. processed_img = np.array(binary_img) #Convert to numpy array #Now save to a tiff file if you want.
@@DigitalSreeni Thanks for the reply. My concern is that is processing images slice by slice may be not the same as using a real 3D filter. For example, Imagej provides both 2D and 3D Gaussian filters. In 3D Gaussian filters, sigma can be assigned to x, y and z. However, in Scikit-image, it seems that only one sigma can be defined. Also, I think 2D watershed can not assigned the same grain in different slices the same label as 3D watershed does. From Scikit-image docstring, its watershed works for both 2D and 3D images but not sure how it works.
Can I know if there are techniques to detect types of noise in an image. Say for a real dataset of satellite images...how do I find out what filters to use?
In almost all cases, you assume that noise is unstructured, which means it is random. This gives you a few choices. I recently gave a keynote presentation on this topic, here is the link. ruclips.net/video/yO15IISXA1Y/видео.html
If you want to clean dead pixels you can use median filter. If you want to detect the location of dead pixels then you need to write code to run median filter and identify hot spots. I do not have anything ready on this topic.
Use glob or os.listdir to read multiple files from a directory and applying a function. I recorded a couple of videos for my work channel, may be they can help. ruclips.net/video/Z90KEqJoC3w/видео.html ruclips.net/video/j6GNtqrwcNE/видео.html
@@DigitalSreeni Sir it gives error when I apply filter to os.listdir(). So can you please give me code for that. Actually I am not able to apply the function to os.listdir
@@DigitalSreeni I have 1 folder which 8 folders in it and each one contains images. Actually I have a multiclass dataset. So I want to apply denoise function to training dataset as well as validation dataset. So please give a code for that.
Sir, can you kindly make a video on 'Anisotropic Defusion Filter' please?? I'm searching for this for a long time but not getting good enough documents or tutorials for python. I would be a huge help and thanks a lot for the tutorials. These helped me a lot in preparing my final year project. :)
I can't thank you enough!! I'm an electrochemist venturing into microscopy and image processing. These are the best introduction to microscopic image processing with python
I am a TEM person and learned a lot from your videos. I am applying what I have learned from you to my research. Thank you for taking the time and effort to make these amazing videos :)
I am applying what I have learned from you to my research. Thank you for taking the time and effort to make these amazing lessons.
Thank you so much sir for making these priceless videos!
I want to do area analysis of particles and phase analysis from an SEM micrograph. I'm facing a problem in separating particles having overlapping areas. Though it identifies individual particles, it also takes their cluster as another separate entity.
Any video on that ?
I'm new to the channel.
Awesome video. Thanks for putting this, and your stuff, out there! Much appreciated. Thank You!
Thank you for such amazing lessons
Ok
Hi what is patch size that you had mentioned in the video? Could you please upload a video describing about it.
Keep it up I'm learning alot from you.
Thanks for sharing. I am looking for filter or denoising grains image as shared in this video example but also has stress release fractures that cut each other and make close boundaries and resulting in misleading grain size…? I hope I explain my question well.
If i want to remove a black coloured lines from my image then, what is best algorithm should be used ?
Thank you for the fantastic tutorials. I have watched tutorial 16 to 26. These tutorials deal with 2D images. My project uses 3D tif. Do the functions in these tutorials also work for 3D images? Do you have a guide about it?
I will record a video on how to work with 3D images. Until then, here is some info...
2D images have numpy array shape as (x, y) for gray images and (x, y, RGB) for color images.
3D images typically have structure (z, x, y) for gray and (z, x, y, RGB) for color image stacks.
Most opencv and scikit-image functions are designed for 2D images. Therefore you need to read one slice at a time inside a for loop and apply the image processing function. Easier than it sounds...
If I have an image of size (50, 256, 256) - 50 slices gray. If I want to apply threshold to the entire volume, slice by slice.
binary_img = []
for img in range(image.shape[0]: #This would be 50 in our example
input_img = image[img, :, :] #Define temporary image slice name
thresh = 75 #Or use otsu
binary = input_img > thresh
binary_img.append(binary) #Add binarized slice to our placeholder list.
processed_img = np.array(binary_img) #Convert to numpy array
#Now save to a tiff file if you want.
@@DigitalSreeni Thanks for the reply. My concern is that is processing images slice by slice may be not the same as using a real 3D filter. For example, Imagej provides both 2D and 3D Gaussian filters. In 3D Gaussian filters, sigma can be assigned to x, y and z. However, in Scikit-image, it seems that only one sigma can be defined. Also, I think 2D watershed can not assigned the same grain in different slices the same label as 3D watershed does. From Scikit-image docstring, its watershed works for both 2D and 3D images but not sure how it works.
Thank you so much, your work is so useful !
You are welcome!
Can I know if there are techniques to detect types of noise in an image. Say for a real dataset of satellite images...how do I find out what filters to use?
In almost all cases, you assume that noise is unstructured, which means it is random. This gives you a few choices. I recently gave a keynote presentation on this topic, here is the link. ruclips.net/video/yO15IISXA1Y/видео.html
@@DigitalSreeni Thank you so much sir
But why do we denoise the images? What is the purpose?
how to check noise in image?
You're amazing, thank you so much!!
Do you have an image processing code to detect the dead pixel in an image ?
If you want to clean dead pixels you can use median filter. If you want to detect the location of dead pixels then you need to write code to run median filter and identify hot spots. I do not have anything ready on this topic.
Sir ,but how i can de-noise the folder which contains 10 noisy images..??? Means my question is how can we de-noise the 10 images at a time.???
Use glob or os.listdir to read multiple files from a directory and applying a function. I recorded a couple of videos for my work channel, may be they can help.
ruclips.net/video/Z90KEqJoC3w/видео.html
ruclips.net/video/j6GNtqrwcNE/видео.html
@@DigitalSreeni
Sir it gives error when I apply filter to os.listdir(). So can you please give me code for that. Actually I am not able to apply the function to os.listdir
@@DigitalSreeni
I have 1 folder which 8 folders in it and each one contains images.
Actually I have a multiclass dataset. So I want to apply denoise function to training dataset as well as validation dataset. So please give a code for that.
Sir, can you kindly make a video on 'Anisotropic Defusion Filter' please??
I'm searching for this for a long time but not getting good enough documents or tutorials for python.
I would be a huge help and thanks a lot for the tutorials. These helped me a lot in preparing my final year project. :)
Did you try medpy library, I remember them having a 2D version of that filter.
thanks sir~!
Good video
Great