To produce a max output from an image

Hi all,

I have a image with 4577x4088 dimension. However, I applied this formula: B(x) = max (Mk(x)) to image, the output I get is in image but dimension became 1x4088. So, how could I remain the 4577x4088 dimension after apply the formula?

[KSSV]

If you pick max from the matrix..there exists only one maximum value..how you are expecting the same dimension again?

[Guillaume]

In addition to what KSSV said,

I applied this formula: B(x) = max (Mk(x))

We have no idea what B, x, Mk are. I suspect that Mk is your image but we certainly have no idea what x is. If x is scalar, then the maximum of a single value is itself and the whole thing is a waste of time.

[Tze Huey Tam]

x is the value of each pixel

[Walter Roberson]

And what is Mk?

[Tze Huey Tam]

Mk indicates the spectral value of pixel x at band k

[Kevin Chng]

Or you may elaborate further what do you want to achieve?

[Tze Huey Tam]

I want to apply Morphological Building Index to image. The first step of this index is to get the maximum value of image.. For more details about this index, please referred to this paper "A Multidirectional and Multiscale Morphological Index for Automatic Building Extraction from Multispectral GeoEye-1 Imagery"

[Walter Roberson]

Unless Mk returns multiple values for every input pixel, then the size of the result of Mk is no bigger than the size of the image. When you apply single input max() to something no bigger than the image, you are going to get a result that is smaller than the image.

There is a two input version of max() as well. max(A, B) is the location by location maximum of the two corresponding locations in the two arrays, and the size of the result is the size of the input array.

For example it is valid to max(A,0) which is equivalent to replacing every negative value in A with 0.

But when there is only one input then the output of max is to take the maximum along the first nonscalar dimension. In your case that would be the maximum of each column.

[Tze Huey Tam]

Thanks Walter...Thanks you very much for your explanation and I get my result. Thank you.

[Walter Roberson]

You can max(max(Mk(x)) to get the max across the whole 2d array