Spectral estimation of fluorescent lamps using RGB digital camera and standard color chart

Spectral estimation of fluorescent lamps is important in practical applications. Using an RGB digital camera and a standard color chart can be regarded as a multispectral way to recover lamp spectral characteristics that contain spikes or spectral lines. In this study, a new method for estimating the spectral power distribution (SPD) of fluorescent lamps was proposed with a workflow by applying two improvements to the existing reflectance recovery methods. The first improvement is to classify the estimated SPD into the corresponding type of fluorescent illuminant defined by CIE based on the spectral correlation distance, and the second one is to choose the optimized number of color patches for lamp SPD recovery with less error. The workflow consists of three modules, i.e. training, testing and optimization procedures. Compared with the traditional recovery method, the training procedure using synthetic data reduces the tremendous workload in practical measurements of different light sources. The testing results from synthetic and real data both indicate that these proposed modifications can significantly improve the accuracy of the existing recovery methods on colorimetric and spectral metrics. And by comparing their experimental results, three among five different methods are suggested to be suitable for the fluorescent lamp spectral estimation.