Hi, I'm trying to convert my electron microscopes for display on a PC rather than the awkward and obsolete consoles. I recently bought an NI IMAQ-1422 (ebay of course) to do the higher level framing and rasterizing. I'll use either a delta-sigma or sampling SAR 16 bit AD converter for the digitizing and a pair of DACS driven by count-down timers to scan the beam. Settling time aside, then, the beam will march along in lock-step with the sampling. Electron microscopes generally use a small, positively charged phosphor screen attached to a light pipe which communicates the light pulses from the screen to a photomultiplier tube. These systems are usually considered to have a nominal bandwidth of 5 Mhz and about 72 dB SNR, though the response time and SNR of more exotic YAP or YAG single crystal scintillators are said to be better. Ultimately I'm shooting for a 512*512 preview image that's updated about 5-10 times per second, then switching to a 1024*1024 image that's updated once each second or thereabouts. I'll probably leave room in the design for at least 4K*4K images to be stored to the HD. But my question is this. Do I have to be concerned about Nyquist considerations such as oversampling and bandwidth limiting for a raster scan application like this? I think I stumbled upon an article somewhere indicating that yes, the same considerations apply. I suppose with the delta-sigma converters I'm looking at, the filtering is built in and there's not much I can do about it or much to worry about. But should I choose an SAR..... I'd "try and see" but I can't find any high performance A/D's in DIP packages. So I'm looking at ~$150 for an EVM and don't want to make a mistake. Perhaps if the sample had periodic features, would I get a Moire type of effect if I didn't Nyquist filter, and would Nyquist filtering eliminate these types of effects? Thanks! -Jeff
Do Nyquist/filtering requirments hold for raster video digitizing?
Started by ●February 10, 2005
Reply by ●February 10, 20052005-02-10
"Jeff Miller" <cornheadorama@hotmail.com> wrote in message news:zWBOd.2724$lz5.323@newssvr24.news.prodigy.net...> Hi, > > I'm trying to convert my electron microscopes for display on a PC rather > than the awkward and obsolete consoles. I recently bought an NI IMAQ-1422 > (ebay of course) to do the higher level framing and rasterizing. I'll use > either a delta-sigma or sampling SAR 16 bit AD converter for the > digitizing and a pair of DACS driven by count-down timers to scan the > beam. Settling time aside, then, the beam will march along in lock-step > with the sampling. > > Electron microscopes generally use a small, positively charged phosphor > screen attached to a light pipe which communicates the light pulses from > the screen to a photomultiplier tube. These systems are usually considered > to have a nominal bandwidth of 5 Mhz and about 72 dB SNR, though the > response time and SNR of more exotic YAP or YAG single crystal > scintillators are said to be better. > > Ultimately I'm shooting for a 512*512 preview image that's updated about > 5-10 times per second, then switching to a 1024*1024 image that's updated > once each second or thereabouts. I'll probably leave room in the design > for at least 4K*4K images to be stored to the HD. > > But my question is this. Do I have to be concerned about Nyquist > considerations such as oversampling and bandwidth limiting for a raster > scan application like this? I think I stumbled upon an article somewhere > indicating that yes, the same considerations apply. I suppose with the > delta-sigma converters I'm looking at, the filtering is built in and > there's not much I can do about it or much to worry about. But should I > choose an SAR..... > > I'd "try and see" but I can't find any high performance A/D's in DIP > packages. So I'm looking at ~$150 for an EVM and don't want to make a > mistake. > > Perhaps if the sample had periodic features, would I get a Moire type of > effect if I didn't Nyquist filter, and would Nyquist filtering eliminate > these types of effects?Yes, you can get a Moire' pattern when there are periodic structures - like picket fences, arrays of windows on large buildings and crystalline structures. That's a very noticeable artifact due to the aliasing. You might consider this: many cameras use discrete arrays for imaging. In that case, the spatial sampling is done immediately and, given that there's no control over the image, there will be aliasing. So, the question is: will the aliasing be a problem for you? Maybe you can know and maybe not.... A classical pattern for testing is a "zone plate". Fred
Reply by ●February 10, 20052005-02-10
Fred Marshall wrote: ...> A classical pattern for testing is a "zone plate".A zone plate has circular symmetry is fine for global tests. The diffraction pattern of a long slit has similar properties but linear form, and so is useful for tests along a particular direction. A zone plate can be thought of as the fringe pattern of a sphere against a flat. Similarly, the pattern whose name I don't recall is the fringe pattern of a cylinder against a flat. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 10, 20052005-02-10
Jerry Avins wrote:> Fred Marshall wrote: > > ... > > >>A classical pattern for testing is a "zone plate". > > > A zone plate has circular symmetry is fine for global tests. The > diffraction pattern of a long slit has similar properties but linear > form, and so is useful for tests along a particular direction. > > A zone plate can be thought of as the fringe pattern of a sphere against > a flat. Similarly, the pattern whose name I don't recall is the fringe > pattern of a cylinder against a flat. > > JerryThanks guys! So... testing issues aside, are you saying I may get moire without oversampling and filtering, and shouldn't get it under any circumstances if I oversample and filter properly? Since I may well be looking at IC's near the resolution limit of the 'scope, it could be an issue. -Jeff
Reply by ●February 10, 20052005-02-10
"jeff miller" <cornheadorama@hotmail.com> wrote in message news:ZPSOd.2990$lz5.477@newssvr24.news.prodigy.net...> Jerry Avins wrote: >> Fred Marshall wrote: >> >> ... >> >> >>>A classical pattern for testing is a "zone plate". >> >> >> A zone plate has circular symmetry is fine for global tests. The >> diffraction pattern of a long slit has similar properties but linear >> form, and so is useful for tests along a particular direction. >> >> A zone plate can be thought of as the fringe pattern of a sphere against >> a flat. Similarly, the pattern whose name I don't recall is the fringe >> pattern of a cylinder against a flat. >> >> Jerry > > Thanks guys! > > So... testing issues aside, are you saying I may get moire without > oversampling and filtering, and shouldn't get it under any circumstances > if I oversample and filter properly? Since I may well be looking at IC's > near the resolution limit of the 'scope, it could be an issue.As before: Yes to your first question. And, that *should* lead directly to a yes to the second question except I'm not so sure about your: "any circumstances". i.e. what if there's a bug? The resolution of the scope could be one measure but it is the most drastic one - "worst case". The spatial frequency content of the images might be less demanding. But only you could know that or hope for it. If the images have higher spatial frequency content than the resolution of the scope will "see", then the scope becomes a filter. If the images have a lower spatial frequency content than the resolution of the scope, then the scope output can be aliased by virtue of subsequent sampling / pizelization. Here's a good example in practice. Let's say we have a very high resolution scanner. Let's say we intend to scan newspaper comics or other pictures. We know that the newspaper creates the pictures with dots using some spatial separaration. We also know that averaging over a number of dots is what's expected of the eye. So, there is a periodic pattern of dots and let's assume it's somewhere around 300dpi. If we scan somewhere around 300dpi, then there will very likely be a Moire' pattern in the output. These can be very evident and annoying. If we scan at very high resolution, then we resolve the dots and get no patterning. If we scan at a much lower resolution, then we average the dots through some mechanism in the scanner that I don't understand. So, the trick is to not scan with a resolution that's similar to the image details. Higher resolution uses more memory. Lower resolution loses some degree of fidelity. In one case, the "camera" is the filter. In the other case, there isn't much of a filter. Where those two cases merge are defined by the original image content. Now let's get back to your sensor / system: If there's a raster scan then you may have already chosen the vertical sampling interval it appears (the number of lines). Having a horizontal sampling interval that's equivalent is at least consistent - that's the sampling interval in space and, in time, along the raster scan lines. It appears the aspect ratio is 1:1 so the same spatial sample interval might be a reasonable choice. Overkill in horizontal is usually easy though and may have some advantage. Just don't forget the vertical resolution embedded in the scan configuration. Fred
Reply by ●February 10, 20052005-02-10
jeff miller wrote:> Jerry Avins wrote: > >> Fred Marshall wrote: >> >> ... >> >> >>> A classical pattern for testing is a "zone plate". >> >> >> >> A zone plate has circular symmetry is fine for global tests. The >> diffraction pattern of a long slit has similar properties but linear >> form, and so is useful for tests along a particular direction. >> >> A zone plate can be thought of as the fringe pattern of a sphere against >> a flat. Similarly, the pattern whose name I don't recall is the fringe >> pattern of a cylinder against a flat. >> >> Jerry > > > Thanks guys! > > So... testing issues aside, are you saying I may get moire without > oversampling and filtering, and shouldn't get it under any circumstances > if I oversample and filter properly? Since I may well be looking at IC's > near the resolution limit of the 'scope, it could be an issue. > > -JeffThat's not how I see it. A digital array samples an image before it can be filtered, so aliasing is not only possible, but likely. One way to reduce the spatial frequency to less than the array spacing is to blur the image, but that's usually counterproductive. Most of the time, you're better off with the artifacts. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 11, 20052005-02-11
Jerry Avins wrote:> jeff miller wrote: > > >>Jerry Avins wrote: >> >> >>>Fred Marshall wrote: >>> >>> ... >>> >>> >>> >>>>A classical pattern for testing is a "zone plate". >>> >>> >>> >>>A zone plate has circular symmetry is fine for global tests. The >>>diffraction pattern of a long slit has similar properties but linear >>>form, and so is useful for tests along a particular direction. >>> >>>A zone plate can be thought of as the fringe pattern of a sphere against >>>a flat. Similarly, the pattern whose name I don't recall is the fringe >>>pattern of a cylinder against a flat. >>> >>>Jerry >> >> >>Thanks guys! >> >>So... testing issues aside, are you saying I may get moire without >>oversampling and filtering, and shouldn't get it under any circumstances >>if I oversample and filter properly? Since I may well be looking at IC's >>near the resolution limit of the 'scope, it could be an issue. >> >>-Jeff > > > That's not how I see it. A digital array samples an image before it can > be filtered, so aliasing is not only possible, but likely. One way to > reduce the spatial frequency to less than the array spacing is to blur > the image, but that's usually counterproductive. Most of the time, > you're better off with the artifacts. > > JerryHmmm, would it be fair to say there's a difference of opinion between Jerry and Fred in these latest responses? The sigma-delta converter I'm looking at has a nominal sample rate of 5Ms/S, but a nominal bandwidth of 2.45Mhz. You don't think that spatial data above 2.45Mhz will simply get tossed, Jerry? Perhaps it's much the same thing as blurring the image. I'm a _little_ disturbed that I might be "wasting" half or 3/4 of my pixels. EM photos are breathtaking because they are so sharp and detailed. I wish I could turn the digital filter on and of. Don't think I can. But for final images, I don't think there's any real limitation on resolution. For display, I could sample at 2Kx2K and dither(?) down to 1Kx1K... though I may need IMAQ Vison (~$700) to do it in real time. For rendering to posters I should be able to capture 32kx32k. Now that Fred mentions it I am aware of "screening" issues when scanning. And sure enough as I understand it the solution is to scan at 2X or more of the screening frequency. I'm leaning toward the sigma-delta converters with built-in filtering so I think I'll be fine in the vertical. But now Fred's got me thinking about the horizontal. How do I "filter" in that dimension? Or oversample for that matter? Perhaps if I can figure out the ultimate resolution (which has mostly to do with "probe size", which as I understand it is approximately equal to the area of the cathode that deos the actual emitting: about 25nm for Tungsten filaments, maybe 10nm for LaB6 cathodes, and 1-2nm for feild emission sources, motly I use Tungsten and LaB6) and then scale my scan for 10nm/pass for Tungsten and 5nm/pass for LaB6... Hmm, well this will be interesting anyway. -Jeff
Reply by ●February 11, 20052005-02-11
"Jeff Miller" <cornheadorama@hotmail.com> wrote in message news:LhaPd.3256$lz5.2965@newssvr24.news.prodigy.net...> Jerry Avins wrote: >> jeff miller wrote: >> >> >>>Jerry Avins wrote: >>> >>> >>>>Fred Marshall wrote: >>>> >>>> ... >>>> >>>> >>>> >>>>>A classical pattern for testing is a "zone plate". >>>> >>>> >>>> >>>>A zone plate has circular symmetry is fine for global tests. The >>>>diffraction pattern of a long slit has similar properties but linear >>>>form, and so is useful for tests along a particular direction. >>>> >>>>A zone plate can be thought of as the fringe pattern of a sphere against >>>>a flat. Similarly, the pattern whose name I don't recall is the fringe >>>>pattern of a cylinder against a flat. >>>> >>>>Jerry >>> >>> >>>Thanks guys! >>> >>>So... testing issues aside, are you saying I may get moire without >>>oversampling and filtering, and shouldn't get it under any circumstances >>>if I oversample and filter properly? Since I may well be looking at IC's >>>near the resolution limit of the 'scope, it could be an issue. >>> >>>-Jeff >> >> >> That's not how I see it. A digital array samples an image before it can >> be filtered, so aliasing is not only possible, but likely. One way to >> reduce the spatial frequency to less than the array spacing is to blur >> the image, but that's usually counterproductive. Most of the time, >> you're better off with the artifacts. >> >> Jerry > > Hmmm, would it be fair to say there's a difference of opinion > between Jerry and Fred in these latest responses? The sigma-delta > converter I'm looking at has a nominal sample rate of 5Ms/S, but a nominal > bandwidth of 2.45Mhz. You don't think that spatial data above 2.45Mhz will > simply get tossed, Jerry? Perhaps it's much the same thing as blurring the > image. > > I'm a _little_ disturbed that I might be "wasting" half or 3/4 of my > pixels. EM photos are breathtaking because they are so sharp and detailed. > I wish I could turn the digital filter on and of. Don't think I can. But > for final images, I don't think there's any real limitation on resolution. > For display, I could sample at 2Kx2K and dither(?) down to 1Kx1K... though > I may need IMAQ Vison (~$700) to do it in real time. For rendering to > posters I should be able to capture 32kx32k. > > Now that Fred mentions it I am aware of "screening" issues when scanning. > And sure enough as I understand it the solution is to scan at 2X or more > of the screening frequency. > > I'm leaning toward the sigma-delta converters with built-in filtering so I > think I'll be fine in the vertical. > > But now Fred's got me thinking about the horizontal. How do I "filter" in > that dimension? Or oversample for that matter? Perhaps if I can figure out > the ultimate resolution (which has mostly to do with "probe size", which > as I understand it is approximately equal to the area of the cathode that > deos the actual emitting: about 25nm for Tungsten filaments, maybe 10nm > for LaB6 cathodes, and 1-2nm for feild emission sources, motly I use > Tungsten and LaB6) and then scale my scan for 10nm/pass for Tungsten and > 5nm/pass for LaB6... > > Hmm, well this will be interesting anyway.Jeff, I don't think that Jerry and I disagree. Probably just looking at different perspectives. I tried to help without actually understanding the arrangement - especially when things are made into discrete pixels - if you will. It appears that you understand the issues pretty well. Let's see if I understand from your description and some of the words used: I have no experience with a scanning electron microscope except to see the results. OK, so it appears that the "scanning" part uses a raster scan mechanism. Of course, there is the possibility that scans can overlap so that oversampling occurs - this is different than video painted on a CRT on on pixelated LCDs - so there may be methods for dealing with this possiblity - I don't know how folks may have done that. Since, in general, you don't know what the image is, then selecting the distance between scan lines is always a trade between performance parameters. So much for vertical. If the horizontal is a continuous scan as it appears, then it will have some "bandwidth" or ability to resolve in space and no more. This limits the bandwidth of the output. If you are going to allow for images that are finer than the sensor will resolve, then the bandwidth of the signal is determined by the sensor. Either you will filter it by blurring the image or by filtering the temporal output of the scan. And then, you can sample without aliasing. As Jerry suggested, it makes sense to consider what you can accomplish by *not* worrying about aliasing. This is very typical in imaging applications I do believe. You take what you can get - even if a Moire' pattern is possible, how likely is it? A picture is worth a lotta words. Let's consider a sampled set of scans: a b c d e f g h i j k 1 o o o o o o o o o o o 2 o o o o o o o o o o o 3 o o o o o o o o o o o 1,2,3 ... are scan lines. The dots in 1 are the samples of 1, etc. a,b,c are the column indices. These are just arrays of numbers and the spatial distance could be anything. So, if 1,2 and 3 are very close together in the vertical then samples a1, a2 and a3 could be virtually identical (oversampled). In your case, it appears that the sensor determines the bandwidth within a scan line. So, sampling a,b,c..... interval might want to theoretically be 1/2 the spatial resolution of the sensor. If that's the case, then making 1,2,3 any closer than that may not gain much. If 1,2,3 are already spaced, then making a,b,c any closer than that distance may not gain much. So, it would be interesting if 1,2,3 spacing isn't somewhat related to the sensor resolution - I have no idea what it is. Fred
Reply by ●February 12, 20052005-02-12
Jeff Miller wrote:> Jerry Avins wrote: > >> jeff miller wrote: >> >> >>> Jerry Avins wrote: >>> >>> >>>> Fred Marshall wrote: >>>> >>>> ... >>>> >>>> >>>> >>>>> A classical pattern for testing is a "zone plate". >>>> >>>> >>>> >>>> >>>> A zone plate has circular symmetry is fine for global tests. The >>>> diffraction pattern of a long slit has similar properties but linear >>>> form, and so is useful for tests along a particular direction. >>>> >>>> A zone plate can be thought of as the fringe pattern of a sphere >>>> against >>>> a flat. Similarly, the pattern whose name I don't recall is the fringe >>>> pattern of a cylinder against a flat. >>>> >>>> Jerry >>> >>> >>> >>> Thanks guys! >>> >>> So... testing issues aside, are you saying I may get moire without >>> oversampling and filtering, and shouldn't get it under any circumstances >>> if I oversample and filter properly? Since I may well be looking at IC's >>> near the resolution limit of the 'scope, it could be an issue. >>> >>> -Jeff >> >> >> >> That's not how I see it. A digital array samples an image before it can >> be filtered, so aliasing is not only possible, but likely. One way to >> reduce the spatial frequency to less than the array spacing is to blur >> the image, but that's usually counterproductive. Most of the time, >> you're better off with the artifacts. >> >> Jerry > > > Hmmm, would it be fair to say there's a difference of opinion > between Jerry and Fred in these latest responses? The sigma-delta > converter I'm looking at has a nominal sample rate of 5Ms/S, but a > nominal bandwidth of 2.45Mhz. You don't think that spatial data above > 2.45Mhz will simply get tossed, Jerry? Perhaps it's much the same thing > as blurring the image. > > I'm a _little_ disturbed that I might be "wasting" half or 3/4 of my > pixels. EM photos are breathtaking because they are so sharp and > detailed. I wish I could turn the digital filter on and of. Don't think > I can. But for final images, I don't think there's any real limitation > on resolution. For display, I could sample at 2Kx2K and dither(?) down > to 1Kx1K... though I may need IMAQ Vison (~$700) to do it in real time. > For rendering to posters I should be able to capture 32kx32k. > > Now that Fred mentions it I am aware of "screening" issues when > scanning. And sure enough as I understand it the solution is to scan at > 2X or more of the screening frequency. > > I'm leaning toward the sigma-delta converters with built-in filtering so > I think I'll be fine in the vertical. > > But now Fred's got me thinking about the horizontal. How do I "filter" > in that dimension? Or oversample for that matter? Perhaps if I can > figure out the ultimate resolution (which has mostly to do with "probe > size", which as I understand it is approximately equal to the area of > the cathode that deos the actual emitting: about 25nm for Tungsten > filaments, maybe 10nm for LaB6 cathodes, and 1-2nm for feild emission > sources, motly I use Tungsten and LaB6) and then scale my scan for > 10nm/pass for Tungsten and 5nm/pass for LaB6... > > Hmm, well this will be interesting anyway. > > -JeffYou betcha! Moire's are infrequent. The only occur when some feature in the image matches the array spacing. The constancy of the line structure of TV rasters allows its removal by a filter. Seeing before and after pictures is one of the experiences that aroused my interest in image processing. Similar results can be had with spot wobble or raised-cosine spot profile, but spatial filtering happens after acquisition. Horizontal filtering is actually easier than vertical. Just low-pass the data stream. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 12, 20052005-02-12
Jeff Miller wrote:> Hmmm, would it be fair to say there's a difference of opinion > between Jerry and Fred in these latest responses? The sigma-delta > converter I'm looking at has a nominal sample rate of 5Ms/S, but a > nominal bandwidth of 2.45Mhz. You don't think that spatial data above > 2.45Mhz will simply get tossed, Jerry? Perhaps it's much the same thing > as blurring the image. > > I'm a _little_ disturbed that I might be "wasting" half or 3/4 of my > pixels. EM photos are breathtaking because they are so sharp and > detailed. I wish I could turn the digital filter on and of. Don't think > I can. But for final images, I don't think there's any real limitation > on resolution. For display, I could sample at 2Kx2K and dither(?) down > to 1Kx1K... though I may need IMAQ Vison (~$700) to do it in real time. > For rendering to posters I should be able to capture 32kx32k. > > Now that Fred mentions it I am aware of "screening" issues when > scanning. And sure enough as I understand it the solution is to scan at > 2X or more of the screening frequency. > > I'm leaning toward the sigma-delta converters with built-in filtering so > I think I'll be fine in the vertical. > > But now Fred's got me thinking about the horizontal. How do I "filter" > in that dimension? Or oversample for that matter? Perhaps if I can > figure out the ultimate resolution (which has mostly to do with "probe > size", which as I understand it is approximately equal to the area of > the cathode that deos the actual emitting: about 25nm for Tungsten > filaments, maybe 10nm for LaB6 cathodes, and 1-2nm for feild emission > sources, motly I use Tungsten and LaB6) and then scale my scan for > 10nm/pass for Tungsten and 5nm/pass for LaB6... > > Hmm, well this will be interesting anyway.Hi Jeff, I haven't been following this before today, but from what I can tell you have not really revealed the necessary information to address the issue of aliasing. If I understand you have a signal that is both digital and analog. In the vertical direction in order to correctly deal with aliasing you need to know how many scan lines are being mapped to how many rows in the array that will store the data. In the horizontal direction that is where the sampling rate comes into play. If you scan 500 lines at 10 times a second you have 5000 lines per second. If your sampler is taking 5 million samples a second then you will have 1000 samples per line. So that is the natural size of the array that your data will fit into 500x1000. If you want to put it into a 512X512 or 1024X1024 or whatever sized array you need to resample using an appropriate resampling scheme. Spline based interpolation works well for image processing resampling - especially if you use tension splines with adjustable tension. That allows you to deal with some of the tradeoffs between aliasing and image clarity on the fly. -jim ----== Posted via Newsfeeds.Com - Unlimited-Uncensored-Secure Usenet News==---- http://www.newsfeeds.com The #1 Newsgroup Service in the World! 120,000+ Newsgroups ----= East and West-Coast Server Farms - Total Privacy via Encryption =----
