ImageReader

[gillins]

Original report by Anonymous.

---

Hi,

Well I had written for a question about use of the function ImageReader, I tried the code in the tutorial to make a unsupervised classification (kmeans) of image of "Lena" using tool ImageReader after I did the classification but with scipy (scipy.ndimage.filters.generic_filter)... and I compared the results, but these are different. why is this happening.

Thanks,
Regards.

[gillins]

Original comment by Neil Flood (Bitbucket: neilflood, GitHub: neilflood).

---

You will need to supply some sort of example code, and a better description of the problem. The information you have given is insufficient for anyone to even understand what is happening, much less to know the possible causes.

[gillins]

Original comment by David Suarez (Bitbucket: GeomatiCol, ).

---

Hi,

Next I send you the code to analise for you.

Thanks, Regards.

[gillins]

Original comment by Neil Flood (Bitbucket: neilflood, GitHub: neilflood).

---

Hi David,

Thanks. I think the main reason for a difference is that the example script you have used filters each band separately. However, using scipy.ndimage.median_filter(img, 3) will filter the three bands together, in 3 dimensions. The equivalent would be something like
filtro = [ndimage.median_filter(img[i], 3) for i in range(3)]
The middle band would be the same, but the first and third bands would be different.

I would like to add that the example you have found (MedianFilterRIOSExample.py) is probably not a very good example. It should almost never be necessary to use these low-level routines in RIOS, such as ImageReader. You should always be able to use applier.apply() to do what you need. I suggest that you only look at the examples which use this.

Thank you for your interest in RIOS.
Neil

[gillins]

Original comment by Neil Flood (Bitbucket: neilflood, GitHub: neilflood).

---

Since I have heard nothing back, I assume that my response was sufficient.