Post by thread:[OT]: Light filters

This is basically a continuation of the auto darkening thread if anybody is interested. I have looked extensively but I really have no idea where to start having had no training in this area. Basically I have a scan head that uses red light to scan the image. What I want to do is make it so that you cannot see the scanhead from outside the casing when it is not in use. I originally thought of some kind of plastic that can have it`s opaque qualities altered electrically, I know think there could be a simpler solution. Would it be possible to use some kind of polarisation filter?? Or possibly a filter that only allows red through. However would this show up as red on the casing, bearing in mind that no light would come from the sensor as it is encased but some could be reflected back from outside. Ideally I am looking for a matt black effect when not in use but as close to that as possible would be great. Any help greatly appreciated.. PS Sorry for the wasted bandwidth I realise this is severely off topic but if it`s any consolation the project has a PIC in it! Andrew Seddon

Andrew Seddon wrote: {Quote hidden}> > This is basically a continuation of the auto darkening thread if anybody is > interested. > > I have looked extensively but I really have no idea where to start having > had no training in this area. Basically I have a scan head that uses red > light to scan the image. What I want to do is make it so that you cannot see > the scanhead from outside the casing when it is not in use. I originally > thought of some kind of plastic that can have it`s opaque qualities altered > electrically, I know think there could be a simpler solution. Would it be > possible to use some kind of polarisation filter?? Or possibly a filter that > only allows red through. However would this show up as red on the casing, > bearing in mind that no light would come from the sensor as it is encased > but some could be reflected back from outside. Ideally I am looking for a > matt black effect when not in use but as close to that as possible would be > great. > > Any help greatly appreciated.. PS Sorry for the wasted bandwidth I realise > this is severely off topic but if it`s any consolation the project has a PIC > in it! > > Andrew Seddon You may want to try translucent red plastic to see what it looks like. It may work good enough. If not, try circular polarized red plastic. The polarization will reduce the reflections that you are worried about. I have used this material for LED displays. This was a long time ago so I don't know where you would get any now. -- Rich

Hi all, Sorry for the off topic. Does anyone know if there are light filters available that will only pass a certain colour band? Eg. yellow only. -- Best regards Tony ICmicro's http://www.picnpoke.com spam_OUTsalesTakeThisOuTpicnpoke.com -- http://www.piclist.com hint: To leave the PICList .....piclist-unsubscribe-requestKILLspam@spam@mitvma.mit.edu

At 02:23 PM 8/25/00 +1000, Tony Nixon wrote: >Hi all, > >Sorry for the off topic. > >Does anyone know if there are light filters available that will only >pass a certain colour band? Eg. yellow only. Kodak, Melles Griot, plenty of places. What color are you looking for? -- Where's dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9 -- http://www.piclist.com hint: To leave the PICList piclist-unsubscribe-requestKILLspammitvma.mit.edu

David VanHorn wrote: > > At 02:23 PM 8/25/00 +1000, Tony Nixon wrote: > >Hi all, > > > >Sorry for the off topic. > > > >Does anyone know if there are light filters available that will only > >pass a certain colour band? Eg. yellow only. > > Kodak, Melles Griot, plenty of places. > > What color are you looking for? I guess now that I think of it, it is a PIC subject. I want to expand on the 'virtual car' project and make it follow a line on the road. A PIC 16F873 now has control over the vehicle speed and manages to get it up to 110KPH and hold it there +/- about 3 KPH. It also mirrors the actions of a dummy steering wheel, but I would like to automate this as well if possible, as it saves me the stress of having to 'drive' it myself. I have done some basic playing around with a video camera to see what signal can be obtained with a yellow line contrasted against a bitumen background. The idea is to use the shift in the yellow signal to decide how far the vehicle is off center and use the steering servo to correct any errors. There are no bends in the line. I was hoping a filter of some sort would help with the contrast by darkening all other colours. -- Best regards Tony ICmicro's http://www.picnpoke.com .....salesKILLspam.....picnpoke.com -- http://www.piclist.com hint: To leave the PICList EraseMEpiclist-unsubscribe-requestspam_OUTTakeThisOuTmitvma.mit.edu

Tony, There are filters that you can aquire to filter colorbands...It depends how sensitive your project is. eg. Some color filters will let for example yellow through but infrared,uv and other more extreme wavelengths will still pass through...If this is no problem for the application at hand then l suggest looking in the RS Catalogue as l think l can remember seeing something there. If not decent camera supply outlets may be able to help. Cheers! {Original Message removed}

Tony Nixon wrote: <snipped> > I have done some basic playing around with a video camera to see what > signal can be obtained with a yellow line contrasted against a bitumen > background. The idea is to use the shift in the yellow signal to decide > how far the vehicle is off center and use the steering servo to correct > any errors. There are no bends in the line. > > I was hoping a filter of some sort would help with the contrast by > darkening all other colours. If you want to "intensify" one particular color, then use a filter of that color. Find a filter of as similar a yellow as possible, and put that on the camera. What the filter does is pass light of the same color as the filter, and attenuate all other colors. Something else to consider is use a black and white camera for this. Putting a yellow filter over the lens of a B&W camera and pointing it at a yellow line will *really* make the line stand out. I'm not exactly sure what bitumen is, but if it's what's generally referred to as asphalt here in the States then it's generally a pretty dark black. And with a B&W camera with a yellow filter, that yellow line outta be damn near impossible to miss. Post-filter, it'll be almost dead white against a dead black background. Of course, this is all presuming that I'm remembering my intro photography course correctly. -- Mike Werner KA8YSD | He that is slow to believe anything and | everything is of great understanding, '91 GS500E | for belief in one false principle is the Morgantown WV | beginning of all unwisdom. -- http://www.piclist.com hint: To leave the PICList piclist-unsubscribe-requestspam_OUTmitvma.mit.edu

At 02:23 PM 8/25/00 +1000, you wrote: Have a look at Edmund Scientific. They should have many such filters. Good luck, Peter {Quote hidden}>Hi all, > >Sorry for the off topic. > >Does anyone know if there are light filters available that will only >pass a certain colour band? Eg. yellow only. > > >-- >Best regards > >Tony > >ICmicro's >http://www.picnpoke.com >@spam@salesKILLspampicnpoke.com > >-- >http://www.piclist.com hint: To leave the PICList >KILLspampiclist-unsubscribe-requestKILLspammitvma.mit.edu > -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestTakeThisOuTmitvma.mit.edu

>Does anyone know if there are light filters available that will only >pass a certain colour band? Eg. yellow only. Try your local theatre suppliers for the cellophane (or whatever i is) they use on stage lighting. A local drama group may even let you have a couple square inches off a piece they have. -- http://www.piclist.com hint: To leave the PICList spamBeGonepiclist-unsubscribe-requestspamBeGonemitvma.mit.edu

Tony said: >I was hoping a filter of some sort would help with the contrast by >darkening all other colours. You've got the right idea. As others pointed out, optimization will mean matching the response of your source/detector/filter combination to the extent possible. Edmund Sci is the most accessible filter vendor in the US. Are you stuck with existing yellow lines, or can you paint your own? A thin line of retroreflective paint and a beamsplitter detection setup would ignore just about everything except the line you want to follow. Contact me off-list if you need further info. Reg Neale -- http://www.piclist.com hint: To leave the PICList TakeThisOuTpiclist-unsubscribe-requestEraseMEspam_OUTmitvma.mit.edu

Tony, instead of painting the road, why don't you use shiny aluminium (duct? air-con?) tape instead? You could do without the cam and go for opto-reflective sensors. Much cheaper to bust up and less complicated. Just a suggestion. Terry {Quote hidden}>I guess now that I think of it, it is a PIC subject. > >I want to expand on the 'virtual car' project and make it follow a line >on the road. A PIC 16F873 now has control over the vehicle speed and >manages to get it up to 110KPH and hold it there +/- about 3 KPH. It >also mirrors the actions of a dummy steering wheel, but I would like to >automate this as well if possible, as it saves me the stress of having >to 'drive' it myself. > >I have done some basic playing around with a video camera to see what >signal can be obtained with a yellow line contrasted against a bitumen >background. The idea is to use the shift in the yellow signal to decide >how far the vehicle is off center and use the steering servo to correct >any errors. There are no bends in the line. > >I was hoping a filter of some sort would help with the contrast by >darkening all other colours. > >-- >Best regards > >Tony > >ICmicro's >http://www.picnpoke.com >RemoveMEsalesTakeThisOuTpicnpoke.com -- http://www.piclist.com hint: To leave the PICList piclist-unsubscribe-requestEraseME.....mitvma.mit.edu

Edmund Scientific sells them. Among other things they carry a filter assortment that gives you a sample of a large number of color filters. The sample set is very reasonably priced, and then you can use it to determine exactly what filter material you want to later order in quantity. We used the filter assortment last school year in a project that determined the color of an object based on the percentage of color measured using a rotating wheel having several color segments. The color assortment was quite good, especially considering the low cost. Fr. Tom McGahee {Original Message removed}

Reginald Neale wrote: {Quote hidden}> > Tony said: > > >I was hoping a filter of some sort would help with the contrast by > >darkening all other colours. > > You've got the right idea. As others pointed out, optimization > will mean matching the response of your source/detector/filter > combination to the extent possible. Edmund Sci is the most > accessible filter vendor in the US. Are you stuck with existing > yellow lines, or can you paint your own? A thin line of > retroreflective paint and a beamsplitter detection setup would > ignore just about everything except the line you want to follow. > Contact me off-list if you need further info. > > Reg Neale > > -- > http://www.piclist.com hint: To leave the PICList > EraseMEpiclist-unsubscribe-requestmitvma.mit.edu A couple of comments: First, you may not be able to get a filter for the exact color of the lines that are painted on the road surface. There are bound to be some variations in the paint. This may mean that you want to use multiple filters to make a wider bandpass to accept more color variation. This is similar in concept to how IF filters are done in television receivers. Second, while riding my bicycle along the white stripe along the shoulder I have noticed that the paint is retroreflective. I think that they sprinkle glass beads onto the paint after applying it to the road. The yellow stripe may be done the same way. I don't plan on riding along the double yellow center line to find out ;-) -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestEraseMEEraseMEmitvma.mit.edu

Richard Ottosen wrote: > First, you may not be able to get a filter for the exact color of the > lines that are painted on the road surface. There are bound to be some > variations in the paint. This may mean that you want to use multiple > filters to make a wider bandpass to accept more color variation. This is > similar in concept to how IF filters are done in television receivers. If you are putting the filters onto the lens of a camera, it doesn't work that way. You will actually be *cutting* the amount of light that gets through drastically, by actually making the effective bandpass narrower. Think of the optical filters as fairly narrow notch filters with roughly equal widths. Putting multiple filters over a camera lens is like putting notch filters with different center frequencies in series. For all intents and purposes you'll be attenuating *everything*. To make up for the variances in the color of the painted line, use a weaker filter. That's the rough equivalent of using a notch filter with a wider pass width. -- Mike Werner KA8YSD | He that is slow to believe anything and | everything is of great understanding, '91 GS500E | for belief in one false principle is the Morgantown WV | beginning of all unwisdom. -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestspam_OUTKILLspammitvma.mit.edu

> Richard Ottosen wrote: > > First, you may not be able to get a filter for the exact color of the > > lines that are painted on the road surface. There are bound to be some > > variations in the paint. This may mean that you want to use multiple > > filters to make a wider bandpass to accept more color variation. This is > > similar in concept to how IF filters are done in television receivers. > > If you are putting the filters onto the lens of a camera, it doesn't work > that way. You will actually be *cutting* the amount of light that gets > through drastically, by actually making the effective bandpass narrower. Even assuming you get filter to do an excellent job, you are going to have to do some image processing to figure out where your line is (and how to correct for it). The amount of processing you need to do seems to require something more than a PIC. You already have WAY more contrast than you need, so you can probably skip the filter. Run the image through an edge detection alg. (soebel, canny, other...) and then filter based on edge orientation (with canny alg.) (since you know that your line is in some orientation..) Group edges into line segments and filter based on some minimum segment length. You can get info on all of this stuff at teh computer vision homepage... http://www.cs.cmu.edu/afs/cs/project/cil/ftp/html/vision.html Or I have code (C++) if it would help. -Steve -- http://www.piclist.com hint: PICList Posts must start with ONE topic: "[PIC]:" PIC only "[EE]:" engineering "[OT]:" off topic "[AD]:" ad's

Go to your local disco and theatrical lighting supply store and ask for for a gell sample book. Voila, about 100 different color plastic filters in 1 X 3" size. Some of them even have spectral response graphs for each sheet. Tony Nixon wrote: {Quote hidden}> Hi all, > > Sorry for the off topic. > > Does anyone know if there are light filters available that will only > pass a certain colour band? Eg. yellow only. > > -- > Best regards > > Tony > > ICmicro's > http://www.picnpoke.com > RemoveMEsalesTakeThisOuTspampicnpoke.com > > -- > http://www.piclist.com hint: To leave the PICList > EraseMEpiclist-unsubscribe-requestspamspamBeGonemitvma.mit.edu -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

>Even assuming you get filter to do an excellent job, you are going to have >to do some image processing to figure out where your line is (and how to >correct for it). The amount of processing you need to do seems to >require something more than a PIC. FYI afaik once upon a time this was done using analog S&H in four zones (two bands with two sides each in the picture) and a small analog computer that computed where the line was 'pointing to' and implemented a 'S' approach to the desired vector using analog PID regulator technology. It all depends on the field of view and the contrast achieved. In theory with 6dB contrast you should be able to get there. A normal B/W camera (that sees the blacktop road and the line only - camera angle down at 60 degrees) will have 35 dB of contrast or more (almost 1:100). A PIC could be programmed to generate the required apertures for S&H and do actual S&H (assuming 16C711). It could output GPS NMEA-like data to simplify the interface to the remaining system. Of course 'real' image processing is more interesting, but do you really want that ? Peter -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

< snip> > I have done some basic playing around with a video camera to see what > signal can be obtained with a yellow line contrasted against a bitumen > background. The idea is to use the shift in the yellow signal to decide > how far the vehicle is off center and use the steering servo to correct > any errors. There are no bends in the line. > > I was hoping a filter of some sort would help with the contrast by > darkening all other colours. > > -- > Best regards > > Tony Have you checked out Glen Williamson's site ? http://www.williamson-labs.com/home.htm Has a section on lane tracking and if you check thoroughly you will find some histograms on line color on asphalt. There's also a graph heading, "color filters' transfer functions." Also: NOT [OT]: Just getting ready to assemble my ROMZAP. Thanks for all you've done for the PIC community Tony. Roger Kadau, netServ -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Two thoughts - How pure is the 'yellow'? Although we may see the paint mix as yellow it probably has a lot of green in it, possibly more green than yellow. Can you take a sample to a hardware store for a (rough) test of the actual colour on their colour matching system? Or to the (instrumental) analytical lab at your uni. The wrong filter could be a disappointing dead end. Narrow wavelength band filters are available - we have a field use (hand held) spectrophotometer for environmental measurement that uses a light globe and a set of narrow band filters to 'select' wavelengths for analysis. The machine is called 'The Analyst' and is made by an American firm. Extech Equipment Pty Ltd, P O Box 579, Boronia, 3155 are the Australian agent, maybe they can help you? Bye. {Original Message removed}

Hi all, Thanks for the tips. I looked at some signals on the CRO from a color CCD camera and "on the surface" it seems like a PIC with an inbuilt comparator should be able to detect the change in light levels in the video signal as long as it is in sync with the framing. These CCD cameras are a lot tougher than I expected. I have had to use two so far out of 7 high speed impact tests. One failed only because the tiny lens cracked. Earlier on, it got pushed through the windscreen by the dash board and landed in the grass about 20 meters away and still worked ok. The PICs are still humming along too. I've just managed to reduce the PIC count from 4 to 2 and shrink 4 prototype boards into a single 4" x 4" PCB. It looks good on paper, but with two servo controllers, (one high power), I can't really judge it until I make one. It's got to be the best Nintendo game I'll ever play :-) -- Best regards Tony ICmicro's http://www.picnpoke.com RemoveMEsalesKILLspampicnpoke.com -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Roger Kadau wrote: > Have you checked out Glen Williamson's site ? > http://www.williamson-labs.com/home.htm > > Has a section on lane tracking and if you check thoroughly you will find > some histograms on line color on asphalt. > > There's also a graph heading, "color filters' transfer functions." Thanks for the tips. > Also: NOT [OT]: > Just getting ready to assemble my ROMZAP. Thanks for all you've done for the > PIC community Tony. > > Roger Kadau, > netServ No probs. I uploaded again only a few days ago because I found a small bug. It's amazing how these things hide until the software gets released. -- Best regards Tony ICmicro's http://www.picnpoke.com salesSTOPspamspam_OUTpicnpoke.com -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

On 25 Aug 2000, at 14:23, Tony Nixon wrote: > Hi all, > > Sorry for the off topic. > > Does anyone know if there are light filters available that will only > pass a certain colour band? Eg. yellow only. > > > -- This will be difficult on a large scale but you could "illuminate" coloured objects with narrow band light sources (the colour of interest) and subtract pre and post illumination intensities. The result is level sensed (digital comparation) and then multiple sample averaged. This should yield very clean data, but the range is intended to be only been 5 metres at no more than 5 kph. I have set up a mono single line CCD sensor but it should be possible with a mono camera. I confess to being stuck on the pattern recognition in the just described method but I am an optimist. _____________________________ Lance Allen Technical Officer Uni of Auckland Psych Dept New Zealand http://www.psych.auckland.ac.nz _____________________________ -- http://www.piclist.com hint: To leave the PICList spamBeGonepiclist-unsubscribe-requestSTOPspamEraseMEmitvma.mit.edu>