There are quite a few questions on SO on image processing that are much more about "how do I detect X" rather than "how do I implement this detection algorithm in language Y", for example this question.

I tend to flag these questions for migration, but it usually takes quite a while till something happens. If the questions already have an accepted answer (like this one), then it may not make a lot of sense to migrate in terms of getting better answers, but they would still help to boost the content on this site (and raise our #of questions rating, as well as possibly traffic).

What should our policy be?


1 Answer 1


What makes a good on-topic question?

I envision this site to be a place for people to bring their research problems (from any field) that involve signal/image processing and get solutions or suggestions for them. There are several inter-disciplinary questions and the recent one on finding stomata in plant microscopy images comes to mind as a good example. The question on river detection in text is yet another great example of someone bringing their problem from a completely different field and asking the experts here if it could be done (and it got great answers!).

Another good use of this site is in understanding algorithms used in certain applications (e.g. algorithm used in photoshop's B & W adjustment layer and explanations on fundamental but often misunderstood concepts (e.g., How are time resolution and frequency bandwidth related). I really, really dislike the "most lucid explanation to teach your grandmother to use wavelets" kind of questions and I hope we don't see them very often. However, that's a different beast to tackle.

What should be our policy?

Ideally, algorithms, ideas and approaches should be first and foremost and not code in language X. I mean, the answerer can use whatever language they feel they're comfortable in (e.g., the river detection question had answers in MATLAB, python and Mathematica), but the focus of the question should not be an explicit "give be the codez in X". Questions like show me how in any language and I'll work it out myself are also most welcome here.

Explicit requests for code in language X should be off-topic. The primary reason for not allowing questions that simply want code in X is because then we'll get overrun by which ever language is most commonly used in the particular field, and other folks who are more inclined towards theory and algorithm development part of it might be put off and leave. I can easily see MATLAB/openCV questions just flooding us if we opened the gates (in fact, we get a few such questions which you don't see because they are either promptly migrated or nuked right here).

Can we have the best of both worlds?

Coming to the question you linked to (electrophoresis gel), I in fact did comment on it and asked the OP if he was looking for code or approaches. He did seem pretty clueless, and looked like he wanted some code to solve his issue right away, so I left it at that. However, barring that, it fits very well with our site's scope and is a good example of the use of image processing in diverse fields. Ultimately, diversity is the key to our growth.

So I decided to try something different. I requested that the question be left on Stack Overflow for a couple of days (a moderator did talk to me about your flag) and once the code part has been solved, I will request migration here, where users can provide an explanation of how to go about it in general. This way, the OP gets his problem solved, and we have a good resource for our site. I realize that our users might not have gotten the head start that Stack Overflow users would've gotten in terms of votes, and so I'm willing to put a bounty on every such question that I request for migration (only new questions), and does not have an already reasonable answer.

I do not know if this will work or if it's a good idea, but I'm willing to give it a try. Since this was less than a day ago, I didn't have the time to bring it here as a meta post, but it's good that you did! I'd appreciate feedback on this idea.

A final word on migrations...

Please do not request for migration of old questions unless if it's really, really off-topic on Stack Overflow (or elsewhere) and a very good question for our site. Borderline and meager questions should be left alone where ever they are. If the OP of the post is not around anymore (greyed out) or hasn't visited in 2 years, please leave them where they are. If there's an old and interesting question but not migration worthy in its current form (either because of poor details in the question or crappy answers), please just re-ask the question here in a better form and in a manner that makes it on-topic for us.

  • $\begingroup$ +1 for a great post, which should go straight into the FAQ. I do, however, disagree re:specific language. I think even if the basic problem of the OP is about finding an algorithm, it is still justifiable to prefer the solution in a specific language. We can make it a policy here that we don't usually care about the language requirements when answering (Belisarius does a great job of that on SO), but I wouldn't want to miss good questions on the site only because an OP stated which language would end up being most useful. $\endgroup$
    – Jonas
    Nov 23, 2011 at 21:42
  • 1
    $\begingroup$ @Jonas That actually might not be a bad idea at all. We can make it such that they can state a preferred language that they'd like the solution to be in, but no guarantees (they can always take the solution and run to SO for a translation :D) and answerers are free to answer in whichever language they choose (or pseudocode). $\endgroup$ Nov 23, 2011 at 21:47
  • $\begingroup$ I think that should work out very well. $\endgroup$
    – Jonas
    Nov 24, 2011 at 16:08

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .