24 - Random Walker segmentation in Python

Histogram based image segmentation is not possible in many cases. This tutorial explains the use of Random Walker segmentation in Python using a noisy backscattered electron (BSE) image collected from an alloy.

[jisoochoi]

18:38 "you don't have any clue, I don't have any clue" What a boss move!!! Thank you so much for posting all of these tutorials. It's a tremendous help for me to get through my project right now!

[nano]

Great explanations.. thank you very much for the tutorial!

[prageethbhanuka]

Great explanation sir! And specially hats off for mentioning the further improvements for denoising using class structure!

[shafagh_projects]

thank you so much for your great tutorial. love them all.

[meghna]

In my console image wouldn't appear.it NumPy boolean array indexing assignment cannot assign 3 input values to the 272685 output values where the mask is true

[kelvinxie]

Very interesting segmentation method. Thank you!

[daljeetsinghranawat]

HELLO, WHAT DOES RANDOM_WALKER DOES TO PIXEL WITH EQ_IMG<0.3????

[ilkingoksal]

Thanks for great explanation, short question here. What if we want to find the average distance between structures by using this contrasts extracted by random walker? should we just find the ratio of one contrast to all area?

[azazkhan]

My ultimate goal is deep learning and training the software to get the images that I want from a huge data. I would like to ask will be able to do so if I follow all the your lectures ?

[Gayatrisomvanshi-hz]

It is great!!! Thank a lot. if we have 3 phases with little lighter difference. then what we can use??

[lorizoli]

As sidenote histogram based segmentation, similar to what was done in the previous lesson with threshold = 0.9 is far from hopeless. It needs a bit of morphological cleanup.

[AhmedEmamAI]

I filled the markers with 1 and 2, but when I show the plt.imshow(markers) I just get black box

[bedangaroymedhi]

The values that you have set after finding the minimum in the histogram would it be applicable for multiple images? As different images might have different minimums. Can you also show an application of Convolutional Neural Networks after applying segmentation.

[XinhLe]

it seems that fast_mode = False is not so bad for me

[arunstartupgur]

Hi, . Very impressive and useful series on image segmentation. I am trying to use it for tree canopy detection on a Google Earth images of a landscape. Can we also do counting of tree canopies with these algorithms? If yes, please share a video on this. U r doing great job man.

[neptne]

Hi!

when I run this:
plt.imshow(all_segments)

this message appeared on the console:
ValueError: NumPy boolean array indexing assignment cannot assign 3 input values to the 4546468 output values where the mask is true

Any idea how to solve it?. Thanks.

[emojiman]

Great video! This solves my problem with DAPI images. But, i can not save the eq_img (created at 11:35) as tif file. Does anyone know how to do it or is it even possible?
Thanks

[sathiyal]

Nice tutorial sir. Sir may I use this code for finger print image segmentation

[niharkashyap]

Can we use this for images with more than 2 segments?