Converting Histogram of Image

I have filtered an image and I have histogram of de-noised image. When I compared this histogram with Reference Image histogram, I noticed that there is a little different between them. How can I convert this histogram to be like Reference Image histogram?

This is hist of Reference Image.

And this is the histogram of de-noised Image:

[Walter Roberson]

That is an odd reference histogram. It looks like the odd locations and even locations were handled differently, with one of them being roughly 2 1/4 times the other.

[Image Analyst]

A denoised image histogram won't necessarily look like the initial noisy image (of course), or the noise-free image. Why do you think it should, unless you've done a nearly perfect job at removing noise. What do your reference, noisy, and denoised images look like? What type of noise do they have and what method did you use to denoise it?

[mohamad mohebian]

My image is a CT and I have used a modified ICA . I got PSNR : 33 but I need more. The noisy image is based on low-dose registeration.

[Image Analyst]

All I can say is that you then need to come up with a better denoising routine.

Keep in mind though that many denoising routines remove what they think are noise which does not necessarily guarantee that they will return to the original image that noise was added to.

Let's say I was looking a a photo of sand, and then I added noise to the sand. It might look nearly identical to the original sand photo. Then removing noise might make the image smoother and that might have less noise in terms of variation overall and in local neighborhoods but that won't necessarily make it more like the original image.

[Walter Roberson]

That up/down/up/down for the histogram does not look plausible to me for CT -- **unless* the histogram is being taken at the wrong width or with badly planned centers, or other similar mistakes are made in processing.

For example if you have uniform random over [0,2] and you round() that, then 1/4 will go to 0, 1/2 will go to 1, and 1/4 will go to 2, whereas most people would probably have guessed that 1/2 would go to 1 and 1/2 would go to 2.

Answer 1

[Rahul punk]

https://www.mathworks.com/matlabcentral/fileexchange/70220-rahulpunk?s_tid=prof_contriblnk