pixel rearrangement in an grayscale image

Hello:

I have an gray scale

image (1620, 1220), this is a noisy speckle image, I would like to rearrange the pixels of this image in a Gaussian fashion, i.e the all the brightest pixels should be put together(near the mean) followed by the less bright pixels away from the mean, can this be done?

[Walter Roberson]

Do you mean that the bright pixels should be put together near the middle? The mean is the average of all of the pixels including the dark ones, and if you put the bright ones together they are going to be brighter than the mean.

[siddharth]

Let me rephrase my question, I want to arrange all the pixels in such a way that all the brightest pixels of same value( on a scale of 0-255) are together, starting from the center of the image(810,610) as the image resolution is 1620 x1220. then the less bright pixels lie on on the periphery and so on. the line profile is should be like, the pixel intensities decreases exponentially as one move radially out from the center(brightest) pixel to the less brighter pixels (in any direction). the analogue of this would be the Gaussian beam profile of a laser, when imaged on a charge coupled device camera.

Answer 1

[Walter Roberson]

Sort the values. Then write them into a matrix of the right size in a spiral order. except that then some pixels of the same value might be spread out in a chain which is one definition of together but a different definition would require clumping...

http://blogs.mathworks.com/videos/2008/09/25/puzzler-cleverness-needed/