bay2rg

Algorithm:

Read the original image.




Separate the original matrix/image in to 3 matrices: R,G,B
Odd Row, Odd Column = B
Odd Row, Even Column = R
Even Row, Odd Column = R
Even Row, Even Column = G

generate---




Each of these 3 new matrices is missing pixel data
(where the other colours used to be) so use kernels
to fill in missing data.
*Note: In order to prevent addition overflow (>255), normalize the
kernel values so their total sum is 1.
* Also note: The kernel is only applied to pixels where data is missing. No need to colour in pixels that are already coloured.

There are 4 possible pixel layouts, so eight 3X3 Kernels are needed:
1)
RGR
BRB
RGR
1b) Kernel to colour centre Blue: {(0,0,0),(0.5,0,0.5),(0,0,0)}
1g) Kernel to colour centre Green: {(0,0.5,0),(0,0,0),(0,0.5,0)}

2)
GRG
RBR
GRG
2r) Kernel to colour centre Red: {(0,0.25,0),(0.25,0,0.25),(0,0.25,0)}
2g) Kernel to colour centre Green: {(0.25,0,0.25),(0,0,0),(0.25,0,0.25)}


3)
BRB
RGR
BRB
3b) Kernel to colour centre Blue: {(0.25,0,0.25),(0,0,0),(0.25,0,0.25)}
3r) Kernel to colour centre Red:  {(0,0.25,0),(0.25,0,0.25),(0,0.25,0)}

4)
RBR
GRG
RBR
4b) Kernel to colour centre Blue: {(0,0.5,0),(0,0,0),(0,0.5,0)}
4g) Kernel to colour centre Green: {(0,0,0),(0.5,0,0.5),(0,0,0)}

Note: Some of these kernels are duplicates, so really only 4 kernels are needed:
i) 1b, 4g -> {(0,0,0),(0.5,0,0.5),(0,0,0)}
ii) 1g, 4b -> {(0,0.5,0),(0,0,0),(0,0.5,0)}
iii) 2g, 3b -> {(0.25,0,0.25),(0,0,0),(0.25,0,0.25)}
iv) 2r, 3r -> {(0,0.25,0),(0.25,0,0.25),(0,0.25,0)}


This will leave you with 8 separate matrices. Combine the Rs in to a single complete R matrix, combine the Bs in to a complete B matrix, the Gs in to a complete G matrix.

Then combine these 3 full matrices in to a single 3-channel color image.

Display.