Adjacent pixels above threshold

[jamier]

Hi All,

Probably a simple question but here goes. I want to find pixels in my image above a threshold of 2. I can do this no problem. However, my image is quite noisy and my threshold image has a lot of noisy pixels surviving. Is there a straightforward method to find unilateral and bilateral occurrences of 2 or 3 adjacent pixels all above my threshold value?


Thanks,
Jamie

[David Webster]

You could apply a 3x3 mean filter to the thresholded image and then
re-threshold (or slice) at desrired number of pixls. Note, you will need to
adjust the actual threshold value used to account for the above threshold
pixel value used in the first image and possibly a multiplier by 9 to
account for averaging

On Sat, Jul 16, 2011 at 10:31 AM, jamier <[hidden email]>wrote:

> Hi All,
>
> Probably a simple question but here goes. I want to find pixels in my image
> above a threshold of 2. I can do this no problem. However, my image is
> quite
> noisy and my threshold image has a lot of noisy pixels surviving. Is there
> a
> straightforward method to find unilateral and bilateral occurrences of 2 or
> 3 adjacent pixels all above my threshold value?
>
>
> Thanks,
> Jamie
>
> --
> View this message in context:
> http://imagej.588099.n2.nabble.com/Adjacent-pixels-above-threshold-tp6590013p6590013.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>

[Michael Schmid]

Hi Jamie,

you can try Process>Filters>Minimum with a small radius (0.5 or 1),  
then Process>Filters>Maximum with the same radius (grayscale  
opening). This will erode small maxima.

Alternatively, you can try Process>Noise>Remove Outliers with a  
threshold of 0 and a small radius.

Michael
________________________________________________________________

On 16 Jul 2011, at 19:31, jamier wrote:

> Hi All,
>
> Probably a simple question but here goes. I want to find pixels in  
> my image
> above a threshold of 2. I can do this no problem. However, my image  
> is quite
> noisy and my threshold image has a lot of noisy pixels surviving.  
> Is there a
> straightforward method to find unilateral and bilateral occurrences  
> of 2 or
> 3 adjacent pixels all above my threshold value?
>
>
> Thanks,
> Jamie