Reconstructing the source modulated camera sensitivity of smartphone from principal component-driven Wiener estimation

Abstract

We reconstructed the camera sensitivity of a smartphone using an indirect estimation method based only on the RGB values of the patches from the captured Macbeth chart. Knowing the camera sensitivity is needed in spectral imaging applications. While a direct estimation method using a diffraction grating and discharge tube is more accurate, it is time-consuming and expensive. Alternatively, indirect estimation method can be solved using Wiener estimation but the reconstructed sensitivity is unreliable. In this study, we incorporated the principal components of the camera sensitivities of 59 cameras – both DSLR and smartphone, on top of the RGB values to estimate the camera sensitivity. Spectral reconstruction results returned sufficient shape similarity and low residual error when using one principal component – making our indirect estimation technique reliable enough to facilitate hyperspectral imaging at home.