image processing- calculate image blur extend and direction

The code was

      c(i)=fft(log(abs(image)));

then calculate c(i') to get the calculate image blur extend and direction. for differentiation of the image i am used sobel filter.bt the diagram(answer) i got was not relevant... i dont know where the mistke? whether using fft or fft2 or fftn for images? or which filter is appropriate for use?

[Merged from duplicate question]

i want to process the image. for the image i have to calculate fft, then abs,then log,t hen again fft. during calculating abs the output was simply white screen. no original image? whats the reason?

[christian bale]

can u be clear with ur question so i can help u

[Walter Roberson]

You should probably be using fft2() for images.

[munch kitty]

k friends..actually in the ieee paper they mentioned some outputs after differntiating the given imagei-i'.for differentiation i used sobel filter and prewitt filter.but the output was not relevant to the output in the paper?where the mistake.whether the filter i used is wrong?

[Walter Roberson]

Umm, IEEE has published at least 4 different papers.

[munch kitty]

sorry sir...i cant get u?4 papers u told.i ask about which filter is used for differentiation?

[Walter Roberson]

You wrote, "actually in the ieee paper" and I am pointing out that IEEE has published multiple papers -- in other words we do not know WHICH paper you are looking at.

Answer 1

[christian bale]

manoj kitty try in scilab it would be easy than matlab

[Ganesh]

Hi manoj!!!!!!How many members in ur project?

[munch kitty]

3 sir..

[Ganesh]

How'd u divide ur project module among with ur project members?

[munch kitty]

where are u from? we divided into 3 modules.each of them doing one module..

[Ganesh]

Am from bangalore man...working in DIIP analyst.

[Ganesh]

Just wanna know so far what u ve implemented in project thats y i asked how many members..

Answer 2

[Walter Roberson]

What datatype are you using for your image? If it was an image file that was read in using imread() then the datatype is probably uint8, unsigned 8 bit integers, in which case abs() would not have any effect.

[munch kitty]

we cant know which data type.but we read the image using imread and process the image.but u told that data type is uint8 or unsigned 8 bit.then how we calculate the abs value for images?

[Walter Roberson]

Well, abs() of an unsigned value is just going to be the same value, so you could skip the abs.

But to double-check: are you sure it was not fft(abs(log(image))) with the abs() applied to the log() ?

[Walter Roberson]

class() will tell you which datatype something is. double, uint8, uint16, and so on.

[munch kitty]

no sir after calculating abs of image and then log sir.so we skip the abs() ? a sir..

[Walter Roberson]

I seem to have missed you showing us the class() of your image.

In your other active thread, you have fft(log(abs(fft(image)))) with an fft() as the lowest level operation. It makes sense to abs() the result of fft, whereas it does not make sense to abs() the image directly.

[munch kitty]

sir did u know about radon tranformation?

[Walter Roberson]

I glanced at the description a number of months ago, but I did not see the point of it at that time, so I am not likely to be able to describe how to use it.