My last thesis update was way back in November '08. I gave the most recent description of my research, and talked about a small bit of theory that I needed to confirm before moving on. Back then (and up until now), my thesis was supposed to be a simple experimentation of a particular class of interest point detectors to show how well they worked for matching photos with cube and cylinder panoramic images.
Unfortunately, I was never quite able to get the photos and panoramas to match automatically in any sort of reasonable way. I tried a few different experiments to show this, but it was difficult to convince ourselves that the strategy used just wasn't working until I really pared it down to the bare essentials. Basically, what I ended up showing was the inadequacy of the interest point descriptors in the context of comparing them in the panoramic and photographic images. You can see some of the results here.
Now I must seek another reasonable method for matching these particular images, given the following:
- The panoramic images are not terribly sharp in their quality. This is related to the fact that some areas of the image are sampled from several sources. Fine features are thus unlikely to be reliably detectable.
- In most comparisons, buildings will be the only structure that will appear in a scene consistently. The ambiguity from repeated structures in these buildings (such as windows) will pose a challenge.
- It may be reasonable to assume that we are always matching images of the same scene thanks to a previously run recognition algorithm and/or GPS location attempt.
- The matching process must result in being able to find the geometry between the cameras of the two scenes.
I am looking into solutions for some of the same problems myself. I'll post if I figure it out.
ReplyDeleteI can really relate! Your predicament reminds me of the philosopher Karl Popper, who said that you are doing true science if your theories are falsifiable. In other words, science looks like this:
ReplyDelete- try something, then it fails
- try something else, then IT fails
- adjust experiment, try something else, then it fails
- etc.!!
Each time you venture down a "wrong path" you are actually learning something, namely, that the falsifiable theory is not the explanation. If your theory is structured such that evidence can only support it, then you are only doing pseudo-science (example: astrology).
So I think that your struggles are a sign that you are doing real, respectable science -- keep it up!!!
We computer scientists have it a little different, I think... rather than "discovering" something in nature, we have to "build" it, THEN evaluate its success. I think we are probably more prone to wrong paths than other scientists.
Just some thoughts!
Frozo
It's too bad that the academic community generally doesn't accept work that shows how something shouldn't be done. On the one hand, this could be a good thing - the literature could be overfilled with this type of paper. Still, there are some cases where it should be accepted so that the community knows what not to do, even if nobody is sure yet what the answer is.
ReplyDeleteHi Gail! I love the new layout of the blog.
ReplyDeleteI feel your pain, with your research difficulties...it really can feel like trying to walk in a darkened room, no flashlight, and lots of junk on the floor. It can be terrifying! But a methodical, non-panicked approach always helps.
I went through a similar thing recently where I started trying to make comparisons at a high level, find out it doesn't seem to work, reduce it down a little, still doesn't seem to work, and I went through multiple iterations of simplifying to problem down in order to get at the crux of the issue. It can take painfully long, because you never know at which level you will find the key. I wish you good luck!!!
And yes, wouldn't it be great if novel debunk papers were just as sexy as novel method papers?
Phizzle, it's nice to hear that you went through the same thing. I do remember how elated you were when you finally got the kick ass results you were hoping for, and so I look forward to that moment ;) So far, I haven't been too discouraged by the change in direction; rather, I have a newfound enthusiasm to spend more time on my thesis research! In any case, I definitely learned to start at the beginning this time around, and so I feel good about that. I'm also going to be smarter about implementation, etc.
ReplyDeleteHi,Gail
ReplyDeleteI once experienced the case before. The results were not satisfying after much effort and I had to change another method. The effort I had payed seemed useless, however, later, I found it could be used for other problems. So any effort you have made is valid sometime.
Don't give it up, I am sure you can find a good method at last.