Replacing value in satellite image.

Dear,

I have a Satellite image 10980x10980. Some value in this image are 9.3886e+36. I want to replace these pixel with value NaN to compute mean of this Image.

I tried to use img(img==9.3886e+36)=NaN but somehow it is not working at all. I try this with a small matrix 2x8 then it is work but with my image it is not replace anything at all. Can anyone help me to solve this?

Best regards,

Answer 1

[Image Analyst]

Use ismembertol():

badPixels = ismembertol(img, 9.3886e+36, 1e36);
img(badPixels) = nan;

[Le Ngo Huy Phan]

Dear Image Analyst,

I have tried ismembertol() like you give. Somehow, all the value changing into NaN. Do you have any idea about this problem? Sorry I cannot attached the image because it is quite large.

Thanks

[Image Analyst]

Try reducing the third argument to get as close as you need to, to the target value, without including all the other values. For example, maybe 1e20 or even 1000.

badPixels = ismembertol(img, 9.3886e+36, 1000);
img(badPixels) = nan;

What are the values of the "good" pixels that you want to keep?

[Le Ngo Huy Phan]

I want the to keep the value <1000 and above that become NaN. I tried few times with this function and see that the badPixels has 2 value which is 1 and 0. 1 mean true, the value that I want to keep and 0 mean false the value that I don't want to keep. However, the pixel have value >1000 is equal 1 in badPixels which mean I have to keep it. Only pixel have value NaN is equal to 0 in badPixels.

[Image Analyst]

That's different. Then just use thresholding to create a mask of logical indexes:

mask = img >= 1000; % Mask of bad pixels to be replaced by nan.
img(mask) = nan; % Do the assignment.

[Le Ngo Huy Phan]

Ah yes, thank you. I still wonder why the img(img>=1000)=NaN not working. So strange tho

[Image Analyst]

That would work too. Just look:

img = randi(2000, 8, 6)
img(img>=1000)=NaN

img =

1946 347 1314 745 539 1333

1298 782 1256 397 846 1079

1601 1663 584 980 1096 1397

908 1607 864 679 1886 1334

865 121 31 1904 836 357

1651 799 1969 1841 1967 257

167 1054 335 106 603 1999

267 834 213 1476 1403 343

img =

NaN 347 NaN 745 539 NaN

NaN 782 NaN 397 846 NaN

NaN NaN 584 980 NaN NaN

908 NaN 864 679 NaN NaN

865 121 31 NaN 836 357

NaN 799 NaN NaN NaN 257

167 NaN 335 106 603 NaN

267 834 213 NaN NaN 343