Computational method for optical spectrum analysis using RGB to Hue transformation

Spectrometers measure optical power as a function of wavelength. Optical spectrum analysis is very important part of optical spectroscopy applications. A low cost DIY spectrometer can be created using DVD-R as diffraction grating and camera as the sensor. Pre-calibration of the DIY spectrometer using a mercury vapor based reference light source is essential. The work discussed in this paper is a proposed methodology wherein the image of emission spectra of light sources can be mapped to the visible wavelength spectrum without any pre-calibration process. In this work RGB triplet in the spectrograph image of a 15W CFL bulb is converted to Hue in the HSV color space and Hue value is mapped from 0^o to 360^o portion of the Hue to 750nm-380nm of the visible spectrum. Results of the study indicate that Hue transformation technique to map RGB pixels to wavelength can be used with an error of +3%. The error to some extent can be minimized by making use of the raw image captured by the camera spectrometer and proper alignment of image on the camera sensor.