Colorize the bw

Ever thought that it is possible to obtain coloured images from a monochrome camera? Colour-enabled camera are nowadays very cheap and it is not an actual functionality to be achieved. On the other side linear cameras, as the one installed in the scanner developed in the SISAR project, are relatively expensive, and the price difference between monochrome and colour-enabled cameras is relevant (from hundreds to thousands €).

In order to extract significant musical information from rolls and cardboards, monochrome images are typically enough, since both punched (holes), printed and hand-written information can be well distinguished. Rare are the cases in which the ink colour is significant. They are rare, but they exist! For examples, in some cardboards for multi-keyboard automatic organs, the ink colour define the keyboard stops activations/deactivations are referred to.

For the (rare) cases in which this is required, we are currently investigating a scanning procedure of the SISAR scanner which allows to obtain coloured images by a monochrome camera.


A coloured image is made up of the composition of the intensity levels of different frequencies in the visible electromagnetic spectrum. Red, Green and Blue are the primary addictive colours (ref) by which, with a good approximation, it is possible to obtain a wide range of colours. Sensitive pixel elements installed in CCD sensors of digital cameras are, exactly for this reason, Red, Green and Blue (hence RGB). Each set of pixel measures the electromagnetic energy for its specific frequency, defining the related image channel. In conclusion, the combination of three channels define what we call a coloured image.

The energy level at certain frequencies can be also obtained properly illuminating the object, not with a white light (ideal white is composed by the maximum energy in all the visible spectrum), but illuminating it with coloured light (e.g. RGB led strips). Instead of filtering the overall spectrum energy at three certain frequencies (as in RGB cameras), it is possible to illuminate the object by coloured light, acquire the energy levels (by a monochrome camera, sensible about constantly in all the spectrum) and combine the three channels to construct the coloured image.

LED lights are particularly suitable for this aim, illuminating the object with a very narrow electromagnetic bandwidth, centred in the Red, Green and Blue frequencies, with the advantages of being extremely affordable and characterized by an extremely short ramp-up time (tenths of nanoseconds). This last feature is extremely important since it allows to acquire a number of shots per second, as linear cameras require.