This is clearly not practical. Aren't you forgetting in-flight entertainment for the pilots? While you're happy and snug in your well fitting seat, the poor pilots are probably nodding off, what with the planes flying themselves and all. Now where's that link I saw a while ago to the video with the stewardess helping the pilots concentrate... Hmm
Well, the RAW's are downloaded to a PC that has much more processing power/memory/storage and then beats jpegs pants off. For one, RAW's typically have 12 bits per pixel (=better dynamic range). For another, jpeg throws away much of the high-frequency content that leads to nice and spiffy sharp images. Most camera pipelines' jpeg encoders suck and are set to low quality for various reasons and you eesentially take a decent resolution image and screw it over.
It's all right for quick snaps, but no use for photographs .
No, your 8 MP fall foliage image does not turn into a 2 MP one "at best".
The interpolation from the sub-sampled Bayer to a full-color image (demoasicking) is very sophisticated. Of course, the fact that the color channels are very highly correlated helps a lot. Let's say you don't sample red at a particular pixel. That doesn't mean you know nothing about the intensity of red at that pixel. There's information about that intensity in the nearest red-sampled pixels.
The best demosaicking algorithms also make use of information about red intensity in the neighboring blue and green pixels. Spatial frequency in the green and blue channels closely matches the spatial frequency in the red channel. Edges, which are a problem in interpolation, also are consistent in all color channels. So there's a lot of information there if one is careful enough to use it. Good reconstruction algorithms use all this and give you amazing results.
David Coffin has made available a neat package called dcraw to interpolate and color-correct RAW images. Look into it. It uses some primitive (as compared to what's current in research literature) demosaicking algorithms, but you'll appreciate how the process works. There's a debian package too.
Are you suggesting we should go in there and somehow instill democracy? That has, after all, worked so well in Iraq, Afghanistan, Vietnam, North Korea, Haiti, Panama, Kosovo, Iran, and India.
India? Umm.. Maybe a little bit of looking before the leaping would be in order here. India is a democracy since 1947 when the British left.
Dropping the astronauts and making it a one way ride for robotic instruments would get the cost down to $ 10 mill/mission or so. The shuttle is more than anything else a political instrument to appeal to the dreams and wonder of the masses. A man in space is so much more saleable than a bunch of lenses and computers. Ah well...
Plus the little matter of highly energetic cosmic rays including solar storms and the heay nucleus inter galatic radiation. It seems in the earth-mars trip one of these nucleii will have passed through each and every cell in the body. Shielding will be heavy and expensive. The cost of an earth-mars manne d mission will probably bankrupt the space programs of all space-capable economies.
I am greatly out of my depth here but I've always wondered if this f followed by an i (or an f followed by an f) issue had an explanation. I've noticed this with TeX documents. It screws up such instances in the documents when going from ps2pdf. IIRC they are called ligatures and are particularly difficult to handle in fonts.
Ah, Well. This was an anecodote recited a loong loong time ago by one of my undergrad profs. I guess it was only a rumor and the problem was the spherical aberration all along.
Thanks for the link. I feel adequately educated for the night:)
A bunch of profs and students at Georgia Tech got involved and were able to get very decent results. There's an oft used image of saturn with a part of the rings showing (I think it's in the Matlab test images set) that they worked on.
IIRC they miscalculated or failed to account for the effect of no gravitation on the glass (which is a fluid) when they fashioned the lens on terra firma. When it was sent up, the optical characteristics changed somewhat and blurred the images. Deblurring solved the problem to some extent.
Umm.. The blur can be estimated from the data. It's not trivial but it is done all the time. There are tonnes of techniques, iterative and otherwise, that given a some prior information (which I am sure there is a fair bit of, given that they have a good idea of the optics and sensors involved) can get you a pretty good estimate of the blur.
Deblurring is essentially a convolution operation anyhow. Its (DE)convolution since they want to reverse the original convolution of the original scene that occured due to convolution with the blur kernel.
If you can esitmate the blur, or let's say, the point spread function (PSF) of the blur, deconvolution is the application of the inverse of the said blur.
This is not always a simple operation. Most real world blur PSFs will not be invertible, or easily so, and the inverse operation will be unstable (lead to "blowing up" of teh function). Conditioning may solve some suc problems.
Iterative techniques are useful in many cases and there are many varied different techniques to do this.
Wiener filters are commonly used. A bunch of adaptive techniques based on Wiener filtering concepts are very effective too.
Indeed. A lucky break for them.
It's an amazing coincidence that I just finished reading "Bad Science : The Short Life and Weird Times of Cold Fusion" by Gary Taubes last night. It's a fast paced, light, extremely well-written book that'll put most thriller/mystery type books to shame. Especially for anybody remotely connected wth academia.
He describes the events leading up to and following the great cold fusion sham in the late 80s and early 90s out of Utah U and BYU and quickly picked up by a bunch of people all around the world.
It seems that 400 time background is peanuts. And that's an understatement. Any fusion reaction would be giving out about 20-40 orders more. Yes, thats 10^(20-40). 400 times is easily accounted for by cosmic radiation unless GREAT precautions are taken in measurement, by noise in the instrumentation.
I have of course not broken tradition and RTFA.
-mp-
Re:Perhaps we should take a lesson from THOREAU...
on
Spammers Sue Spamee
·
· Score: 1
Offtopic: I am constantly surprised by the number of people who can't seem to get poor guy's name right. It's spelt M.K. Gandhi. The Ghandi spelling is all over the place for some reason. Probbaly got to be connected to phonetics. 'Ghandi' very closely resembles a colloquial Hindi word for the rear-end.:-)
Incidentally, his doctrine of non-violence is recognized among a large group of people as more of a shrewd political strategy, than something based on an absolute moral framework.
The president of the country is a missile scientist and has been involved in many of the most significant adavances in the Indian missile/nuclear
field. http://www.geocities.com/siafdu/kalam.html
Although, in the Indian system, the president has few executive responsibilities and powers (the prime minister is the head of government), the amount of respect bestowed upon the man is a sign that scientific accomplishment is held in high regard in the country.
Little wonder that this mission and other high technology endeavors do get their well deserved allotment of resources.
-mp-
Slashdot Mirror
This is clearly not practical. Aren't you forgetting in-flight entertainment for the pilots? While you're happy and snug in your well fitting seat, the poor pilots are probably nodding off, what with the planes flying themselves and all. Now where's that link I saw a while ago to the video with the stewardess helping the pilots concentrate... Hmm
It's all right for quick snaps, but no use for photographs .
Look at David Coffin's dcraw. There's a debian package if you are so inclined.
The interpolation from the sub-sampled Bayer to a full-color image (demoasicking) is very sophisticated. Of course, the fact that the color channels are very highly correlated helps a lot. Let's say you don't sample red at a particular pixel. That doesn't mean you know nothing about the intensity of red at that pixel. There's information about that intensity in the nearest red-sampled pixels.
The best demosaicking algorithms also make use of information about red intensity in the neighboring blue and green pixels. Spatial frequency in the green and blue channels closely matches the spatial frequency in the red channel. Edges, which are a problem in interpolation, also are consistent in all color channels. So there's a lot of information there if one is careful enough to use it. Good reconstruction algorithms use all this and give you amazing results. David Coffin has made available a neat package called dcraw to interpolate and color-correct RAW images. Look into it. It uses some primitive (as compared to what's current in research literature) demosaicking algorithms, but you'll appreciate how the process works. There's a debian package too.
Are you suggesting we should go in there and somehow instill democracy? That has, after all, worked so well in Iraq, Afghanistan, Vietnam, North Korea, Haiti, Panama, Kosovo, Iran, and India. India? Umm.. Maybe a little bit of looking before the leaping would be in order here. India is a democracy since 1947 when the British left.
Dropping the astronauts and making it a one way ride for robotic instruments would get the cost down to $ 10 mill/mission or so. The shuttle is more than anything else a political instrument to appeal to the dreams and wonder of the masses. A man in space is so much more saleable than a bunch of lenses and computers. Ah well...
Plus the little matter of highly energetic cosmic rays including solar storms and the heay nucleus inter galatic radiation. It seems in the earth-mars trip one of these nucleii will have passed through each and every cell in the body. Shielding will be heavy and expensive. The cost of an earth-mars manne d mission will probably bankrupt the space programs of all space-capable economies.
I am greatly out of my depth here but I've always wondered if this f followed by an i (or an f followed by an f) issue had an explanation. I've noticed this with TeX documents. It screws up such instances in the documents when going from ps2pdf. IIRC they are called ligatures and are particularly difficult to handle in fonts.
Ah, Well. This was an anecodote recited a loong loong time ago by one of my undergrad profs. I guess it was only a rumor and the problem was the spherical aberration all along. :)
Thanks for the link. I feel adequately educated for the night
A bunch of profs and students at Georgia Tech got involved and were able to get very decent results. There's an oft used image of saturn with a part of the rings showing (I think it's in the Matlab test images set) that they worked on.
IIRC they miscalculated or failed to account for the effect of no gravitation on the glass (which is a fluid) when they fashioned the lens on terra firma. When it was sent up, the optical characteristics changed somewhat and blurred the images. Deblurring solved the problem to some extent.
Umm.. The blur can be estimated from the data. It's not trivial but it is done all the time. There are tonnes of techniques, iterative and otherwise, that given a some prior information (which I am sure there is a fair bit of, given that they have a good idea of the optics and sensors involved) can get you a pretty good estimate of the blur.
Deblurring is essentially a convolution operation anyhow. Its (DE)convolution since they want to reverse the original convolution of the original scene that occured due to convolution with the blur kernel.
If you can esitmate the blur, or let's say, the point spread function (PSF) of the blur, deconvolution is the application of the inverse of the said blur.
This is not always a simple operation. Most real world blur PSFs will not be invertible, or easily so, and the inverse operation will be unstable (lead to "blowing up" of teh function). Conditioning may solve some suc problems.
Iterative techniques are useful in many cases and there are many varied different techniques to do this.
Wiener filters are commonly used. A bunch of adaptive techniques based on Wiener filtering concepts are very effective too.
Indeed. A lucky break for them. It's an amazing coincidence that I just finished reading "Bad Science : The Short Life and Weird Times of Cold Fusion" by Gary Taubes last night. It's a fast paced, light, extremely well-written book that'll put most thriller/mystery type books to shame. Especially for anybody remotely connected wth academia. He describes the events leading up to and following the great cold fusion sham in the late 80s and early 90s out of Utah U and BYU and quickly picked up by a bunch of people all around the world. It seems that 400 time background is peanuts. And that's an understatement. Any fusion reaction would be giving out about 20-40 orders more. Yes, thats 10^(20-40). 400 times is easily accounted for by cosmic radiation unless GREAT precautions are taken in measurement, by noise in the instrumentation. I have of course not broken tradition and RTFA. -mp-
Offtopic: I am constantly surprised by the number of people who can't seem to get poor guy's name right. It's spelt M.K. Gandhi. The Ghandi spelling is all over the place for some reason. Probbaly got to be connected to phonetics. 'Ghandi' very closely resembles a colloquial Hindi word for the rear-end. :-)
Incidentally, his doctrine of non-violence is recognized among a large group of people as more of a shrewd political strategy, than something based on an absolute moral framework.
The president of the country is a missile scientist and has been involved in many of the most significant adavances in the Indian missile/nuclear field. http://www.geocities.com/siafdu/kalam.html Although, in the Indian system, the president has few executive responsibilities and powers (the prime minister is the head of government), the amount of respect bestowed upon the man is a sign that scientific accomplishment is held in high regard in the country. Little wonder that this mission and other high technology endeavors do get their well deserved allotment of resources. -mp-