120 - Image registration methods in python

[@chandrikayadavk1736]

ERROR: No matching distribution found for image_registration.I am getting this error while doing Pip image_registration

[@hayatlr3000]

I'm having the same problem did you fixed it ?

[@kaushalyasivayogaraj5862]

Your image registration videos are highly understandable. Can you make videos on 3D image registration ?. that will be great help.

[@emmanuelschmulewitz1215]

Really good video! Exactly what I was looking for. I tried the third approach but correcting using flow method delivered no convincing result: Is it wrong to use the flow array like this inside the shift function: corrected_image = shift(offset_image,shift=(flow[1],flow[0]),mode='constant') # in x,y (where flow[1] is in x and flow[0] goes into the y direction? Maybe it is obvious but still don't get it...

[@taherehh851]

I got this error ImportError: erfa version 1.7 or later must be installed to use Astropy. erfa is not installed. I installed pyerfa but I have the same problem. Any advice?

[@cltee6629]

hi sreeni, do you have any examples of compare a test image against a golden image and then highlight where the differences are? tq

[@DigitalSreeni]

No, I do not, May be you need to look into anomaly detection in images.

[@borekstvorek]

there are 8 BSE images in the Image folder. however, none of them is called "BSE.jpg" you're not mentioning this and I keep getting error: ValueError: Images must have same shape. any advice will be appreciated as to clear out what images you're using and where exactly to find them

[@DigitalSreeni]

I am not sure why calling image with a specific name makes any difference. The error you are getting indicates that the images you are trying to register are not of the same size (in dimensions).

[@borekstvorek]

@@DigitalSreeni the name is important because there is a discrepancy between the files provided for this tutorial and files being used in this tutorial. Which causes this error. The image said to be used is not present. I got rid of it by trying the other images and the image called BSE.jpg referred to in this tutorial is the BSE_google.jpg in the authors github repo. That one has the same shape/dimensions.

[@ela_bd]

hello, i hope you were well, i have a question about medical image registration. my images are very different from each other and i dont know Is it possible to register them at all. can i send a sample for you and you check them?

[@jerika4005]

Hello , what are your data and what type of registration you need to do ?

[@ela_bd]

@@jerika4005 my images are about cardiac MRI and because of high diffeneces between images i need non rigid transforms.

[@michael_d2]

what would you recommend for registration of scaling?

[@DigitalSreeni]

You may want to investigate elastic registration methods.

[@themaryamsadeghi]

Hi Sreeni, thank you for your great video series. Do you have any recommendations for non-rigid registration methods in python?

[@DigitalSreeni]

Please wathc the next video: 121 - Image registration using pystackreg library in Python

[@themaryamsadeghi]

@@DigitalSreeni Thank you for you response. I watched it now, but I'm looking for a deformable solution, which is not included in that library.

[@mahtabbigverdi3393]

can we apply these methods on 3D images like CT scans with multiple slices? (+ thanks for your great videos)

[@DigitalSreeni]

Yes of course you can apply these methods, especially pystackreg to 3D stacks, like CT images. But it may not make sense for CT images as those are collected / reconstructed as 3D volumes so there will be no mismatch between individual slices.

[@thubuntu]

xoff, yoff, exoff, eyoff = chi2_shift(image, offset_image, noise, return_error=True, upsample_factor='auto') I always recieve following error in VS Code "Exception has occurred: ValueError too many values to unpack (expected 2) File "/home/john/Documents/DSM/testeMal/python_Registrierung/Untitled-1.py", line 23, in shift, error, diffphase = chi2_shift(image, offset_image, noise, return_error=True, upsample_factor='auto') ValueError: too many values to unpack (expected 2)"

[@samarafroz9852]

Nice video sir