How to apply bilateral filter

[alexderjuengere]

---

@gptshubham595 wrote:

Hey Rahul you can use this directly to do this using above python code and python integration toolbox via Enthought 

 

---

 if you don't put 

hl = int(diameter/2)

you'll get 

gi = gaussian(source[neighbour_x, neighbour_y] - source[x,y], sigma=sigma_i)

IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices

and you should add 

 gi = gaussian(x=float(source[neighbour_x, neighbour_y]) - source[x,y], sigma=sigma_i)

or there will be 

gi = gaussian(source[neighbour_x, neighbour_y] - source[x,y], sigma=sigma_i)

RuntimeWarning: overflow encountered in ubyte_scalars

[alexderjuengere]

@gptshubham595 wrote:

.. using above python code and python integration toolbox via Enthought 

rather use the python node

[gptshubham595]

thank you!

ISSUE: ERR IN READING INPUT IMG

I tried your code it is giving me an output image but in order to read input image it is giving error

but can you help regarding this error? 

[gptshubham595]

Sorry it was error due to png file read as jpg file

replacing read jpg file.vi to read png file.vi  worked

ThankYou

[alexderjuengere]

I wanted to see, if I can reduce JPG artefacts by bilateral filtering 

Actually, we don't need the opencv-python package to read jg or png.

attached is  a more polished version, in LabView 2020 x64

[altenbach]

Is the image b&w (8bit greyscale) or RGB (24bit)?

I think all you need to do is a 2D convolution with a 2D Gaussian. Code should fit a postage stamp. No Python.

[altenbach]

---

@altenbach wrote:

Is the image b&w (8bit greyscale) or RGB (24bit)?

I think all you need to do is a 2D convolution with a 2D Gaussian.

---

Here's how that could look like (note that the edges need to be corrected for the truncated kernel, shown is one possible simple correction). Note that the blur scale is arbitrary and might not correspond to other libraries.

[gptshubham595]

the main intention is not to get a blurred image, instead, we wanted to apply a bilateral filter over the image, in order to get a clean curve and reduce noise

like

initially we have (good freq)>=(noise)

frequency*1000 then we have (good freq)_new >> (noise) _new

GKalliatakis/Bilateral-Filtering: Applying Bilateral Filtering to images (github.com)

[altenbach]

Sure. I am just saying that all that could be implemented in G directly. Probably not that hard.

Do you have an example image before and after the desired filtering?

[gptshubham595]