Post by thread:[OT] RGB from bayer mask

Hi guys, Does anyone here know how to get RGB from a Bayer mask image? Given the array bellow, 1 2 3 4 5 6 1 G B G B G B 2 R G R G R G 3 G B G B G B 4 R G R G R G 5 G B G B G B 6 R G R G R G How do you get this? 1 2 3 4 5 6 1 RGB RGB RGB RGB RGB RGB 2 RGB RGB RGB RGB RGB RGB 3 RGB RGB RGB RGB RGB RGB 4 RGB RGB RGB RGB RGB RGB 5 RGB RGB RGB RGB RGB RGB 6 RGB RGB RGB RGB RGB RGB I've tried google but I haven't been able to find anything yet. Thanks. -Mario

Mario Mendes Jr. wrote: {Quote hidden}> Hi guys, > > Does anyone here know how to get RGB from a Bayer mask image? Given the > array bellow, > > 1 2 3 4 5 6 > 1 G B G B G B > 2 R G R G R G > 3 G B G B G B > 4 R G R G R G > 5 G B G B G B > 6 R G R G R G > > How do you get this? > > 1 2 3 4 5 6 > 1 RGB RGB RGB RGB RGB RGB > 2 RGB RGB RGB RGB RGB RGB > 3 RGB RGB RGB RGB RGB RGB > 4 RGB RGB RGB RGB RGB RGB > 5 RGB RGB RGB RGB RGB RGB > 6 RGB RGB RGB RGB RGB RGB > > I've tried google but I haven't been able to find anything yet. > > Thanks. > > > -Mario Mario, You ask the most interesting questions! Here is what I found on Wikipedia: http://en.wikipedia.org/wiki/Calculating_RGB_values Does this article show the way or ???

Mario Mendes Jr. wrote: {Quote hidden}>Hi guys, > >Does anyone here know how to get RGB from a Bayer mask image? Given the >array bellow, > > 1 2 3 4 5 6 >1 G B G B G B >2 R G R G R G >3 G B G B G B >4 R G R G R G >5 G B G B G B >6 R G R G R G > >How do you get this? > > 1 2 3 4 5 6 >1 RGB RGB RGB RGB RGB RGB >2 RGB RGB RGB RGB RGB RGB >3 RGB RGB RGB RGB RGB RGB >4 RGB RGB RGB RGB RGB RGB >5 RGB RGB RGB RGB RGB RGB >6 RGB RGB RGB RGB RGB RGB > >I've tried google but I haven't been able to find anything yet. > > > Have you seen this? http://www-ise.stanford.edu/~tingchen/main.htm Gaston

"You ask the most interesting questions!" Hahahahaha, sometimes I don't know if that is good or bad. Good because I get to learn a little more, but bad be cause now I feel compeled to spend some money on another pic project that will cost 3 times as much as if I were to by a web cam at the store =) Anyway, I found this other site that has a nice visual description (scroll half way down the page) of the process along with some more info on white balancing and the like: http://www.siliconimaging.com/RGB%20Bayer.htm Thanks. -Mario {Original Message removed}

Very nice. Thanks. -Mario {Quote hidden}>Have you seen this? http://www-ise.stanford.edu/~tingchen/main.htm >Gaston >Mario Mendes Jr. wrote: > >>Hi guys, >> >>Does anyone here know how to get RGB from a Bayer mask image? Given >>the array bellow, >> >> 1 2 3 4 5 6 >>1 G B G B G B >>2 R G R G R G >>3 G B G B G B >>4 R G R G R G >>5 G B G B G B >>6 R G R G R G >> >>How do you get this? >> >> 1 2 3 4 5 6 >>1 RGB RGB RGB RGB RGB RGB >>2 RGB RGB RGB RGB RGB RGB >>3 RGB RGB RGB RGB RGB RGB >>4 RGB RGB RGB RGB RGB RGB >>5 RGB RGB RGB RGB RGB RGB >>6 RGB RGB RGB RGB RGB RGB >> >>I've tried google but I haven't been able to find anything yet. >> >> >>

Mario Mendes Jr. wrote: > Does anyone here know how to get RGB from a Bayer mask image? Given the > array bellow, > > 1 2 3 4 5 6 > 1 G B G B G B > 2 R G R G R G > 3 G B G B G B > 4 R G R G R G > 5 G B G B G B > 6 R G R G R G I've never heard of a "Bayer" mask, but why can't you just interpolate the missing components? This sounds like the obvious brute force way. Have you tried it? When green is not present, it's the average of the 4 surrounding green pixels. For red and blue, it's the average of either the two above/below or left/right or the four corners. There are probably more clever ways, but this is to trivial I would try it first and see if it's good enough. ****************************************************************** Embed Inc, Littleton Massachusetts, (978) 742-9014. #1 PIC consultant in 2004 program year. http://www.embedinc.com/products

Bayer grid is usually what you get from CCD/CMOS sensors. All digital Cameras interpolate from that format to RGB/YUV. Usually the equation is a second-degree one (at least) so that you weigh the proximal few layers of the missing color by distance (like a gaussian filter, sort of) to get the missing components. - Marcel Olin Lathrop wrote: {Quote hidden}> Mario Mendes Jr. wrote: > >> Does anyone here know how to get RGB from a Bayer mask image? Given the >> array bellow, >> >> 1 2 3 4 5 6 >> 1 G B G B G B >> 2 R G R G R G >> 3 G B G B G B >> 4 R G R G R G >> 5 G B G B G B >> 6 R G R G R G > > > I've never heard of a "Bayer" mask, but why can't you just interpolate the > missing components? This sounds like the obvious brute force way. Have > you > tried it? > > When green is not present, it's the average of the 4 surrounding green > pixels. For red and blue, it's the average of either the two > above/below or > left/right or the four corners. There are probably more clever ways, but > this is to trivial I would try it first and see if it's good enough. > > > ****************************************************************** > Embed Inc, Littleton Massachusetts, (978) 742-9014. #1 PIC > consultant in 2004 program year. http://www.embedinc.com/products

Marcel Birthelmer wrote: > Bayer grid is usually what you get from CCD/CMOS sensors. All digital > Cameras interpolate from that format to RGB/YUV. Usually the equation is > a second-degree one (at least) so that you weigh the proximal few > layers of the missing color by distance (like a gaussian filter, sort > of) to get the missing components. That makes sense when you want to resample independently of the original pixel grid. However in the case show if you just want the full RGB at the existing pixel locations, the 3rd order fit will most likely reduce to what I suggested. At most you will have very weak contributions from farther pixels, but I doubt that's a good idea. ****************************************************************** Embed Inc, Littleton Massachusetts, (978) 742-9014. #1 PIC consultant in 2004 program year. http://www.embedinc.com/products

On Fri, 2 Dec 2005, Marcel Duchamp wrote: {Quote hidden}> Mario Mendes Jr. wrote: >> Hi guys, >> >> Does anyone here know how to get RGB from a Bayer mask image? Given the >> array bellow, >> >> 1 2 3 4 5 6 >> 1 G B G B G B >> 2 R G R G R G >> 3 G B G B G B >> 4 R G R G R G >> 5 G B G B G B >> 6 R G R G R G >> >> How do you get this? >> >> 1 2 3 4 5 6 >> 1 RGB RGB RGB RGB RGB RGB >> 2 RGB RGB RGB RGB RGB RGB >> 3 RGB RGB RGB RGB RGB RGB >> 4 RGB RGB RGB RGB RGB RGB >> 5 RGB RGB RGB RGB RGB RGB >> 6 RGB RGB RGB RGB RGB RGB Notice that there are N G values, N/2 R values and N/2 B values. Thus you can produce a G value for each 2nd pixel and a R and B value for each 3rd pixel. This is the price to pay for not using 3 image pickups (as pro cameras do). If yuu squint you can see that you get two G's and one each R and B from each 'square' formed by pixels 0,0 0,1 1,0 1,1 etc. There are several ways to do it but a simple way is to divide the matrix into squares of 4 pixels and then: row0: Or(0,0) = I(1,0) Og(0,0) = I(0,0) Ob(0,0) = I(0,1) Or(0,1) = I(1,0) Og(0,1) = I(1,1) Ob(0,1) = I(0,1) Or(0,2) = I(1,2) Og(0,2) = I(0,2) Ob(0,2) = I(0,3) Or(0,3) = I(1,2) Og(0,3) = I(1,3) Ob(0,3) = I(0,3) ... On row 1 you use info from row 2 and so on. It's a hack. Later you can use the bilinear interpolation aluded to in the wikipedia article. Peter

On Sat, 3 Dec 2005, Olin Lathrop wrote: {Quote hidden}> Mario Mendes Jr. wrote: >> Does anyone here know how to get RGB from a Bayer mask image? Given the >> array bellow, >> >> 1 2 3 4 5 6 >> 1 G B G B G B >> 2 R G R G R G >> 3 G B G B G B >> 4 R G R G R G >> 5 G B G B G B >> 6 R G R G R G > > I've never heard of a "Bayer" mask, but why can't you just interpolate the > missing components? This sounds like the obvious brute force way. Have you > tried it? > > When green is not present, it's the average of the 4 surrounding green > pixels. For red and blue, it's the average of either the two above/below or > left/right or the four corners. There are probably more clever ways, but > this is to trivial I would try it first and see if it's good enough. The original idea behind the Bayer filter was to get a signal that could lead to the reconstruction of TV Y,R-Y,B-Y signals with simple *analog* means in 1950's or so. That is why it is organized like that (the other reason is that the filter is made by masking stripes in 3 colors onto a substrate - much like stripe crts are better than delta crts). The original extraction required frequency domain filtering (of the signal coming from a vidicon). The G signal would have maximum frequency and it would be modulated by the frequency of the stripes (as scanned horizontally). Then a simple PLL locked to this frequency (or in ultra cheap designs just peaking and limiting) and one H period delay line would be enough to recover R-Y and B-Y using 2 synchronous modulators driven by the G high frequency signal. In a CCD camera the PLL is not necessary as the pixel clock is well known and derived from the main timing chain. Peter