Multispectral imaging with a modified DSLR camera

This paper describes a multispectral imaging system using a Canon EOS 1000D camera, photographic lenses, and a dispersive element. The system consists of three subsystems: a PVC holder to arrange the optical devices that can be connected to the camera, an acrylic frame to support the position of the PVC holder, and expectation maximization algorithm to reconstruct the spectral-spatial data cube. We perform a calibration process to output the reference matrix of the reconstruction using four LEDs with different wavelengths. The output of the system is a data cube in the form of 20 images that represent the observed intensity response from 455 nm to 645 nm.